Advertisement

Graduate Student Literature Review: Detecting health disorders using data from automatic milking systems and associated technologies

Open ArchivePublished:June 27, 2018DOI:https://doi.org/10.3168/jds.2018-14521

      ABSTRACT

      This review synthesizes a range of research findings regarding behavioral and production responses to health disorders and subsequent illness detection for herds using automatic (robotic) milking systems (AMS). We discuss the effects of health disorders on cow behavior and production, specifically those variables that are routinely recorded by AMS and associated technologies. This information is used to inform the resultant use of behavior and production variables and to summarize and critique current illness detection studies. For conventional and AMS herds separately, we examined research from the past 20 yr and those variables recorded automatically on-farm that may respond to development of illness and lameness. The main variables identified were milk yield, rumination time, activity, and body weight, in addition to frequency of successful, refused, and fetched (involuntary) milkings in AMS herds. Whether making comparisons within cow or between sick and healthy cows, consistent reductions in activity, rumination time, and milk yield are observed. Lameness, however, had obvious negative effects on milk yield but not necessarily on rumination time or activity. Finally, we discuss detection models for identifying lameness and other health disorders using routinely collected data in AMS, specifically focusing on their scientific validation and any study limitations that create a need for further research. Of the current studies that have worked toward disease detection, many data have been excluded or separated for isolated models (i.e., fresh cows, certain lactation groups, and cows with multiple illnesses or moderate cases). Thus, future studies should (1) incorporate the entire lactating herd while accounting for stage of lactation and parity of each animal; (2) evaluate the deviations that cows exhibit from their own baseline trajectories and relative to healthy contemporaries; (3) combine the use of several variables into health alerts; and (4) differentiate the probable type of health disorder. Most importantly, no model or software currently exists to integrate data and directly support decision-making, which requires further research to bridge the gap between technology and herd health management.

      Key words

      INTRODUCTION

      Rapid adoption of automatic (robotic) milking systems (AMS) for dairy cows is occurring worldwide. As of 2014, over 25,000 farms globally were using AMS (
      • Barkema H.W.
      • von Keyserlingk M.A.G.
      • Kastelic J.P.
      • Lam T.J.G.M.
      • Luby C.
      • Roy J.-P.
      • LeBlanc S.J.
      • Keefe G.P.
      • Kelton D.F.
      Invited review: Changes in the dairy industry affecting dairy cattle health and welfare.
      ) and this number continues to grow. In Europe, this has been predominantly driven by growth in the Netherlands and Nordic countries, and in North America, Canada is the major domain of AMS use because of stable milk prices through supply management (
      • Barkema H.W.
      • von Keyserlingk M.A.G.
      • Kastelic J.P.
      • Lam T.J.G.M.
      • Luby C.
      • Roy J.-P.
      • LeBlanc S.J.
      • Keefe G.P.
      • Kelton D.F.
      Invited review: Changes in the dairy industry affecting dairy cattle health and welfare.
      ). Benefits of AMS for farmers include reduced labor requirements and greater time flexibility, while cows benefit by having more freedom to control their time budgets (
      • Jacobs J.A.
      • Siegford J.M.
      Invited review: The impact of automatic milking systems on dairy cow management, behavior, health, and welfare.
      ). Regarding health management, a key advantage of AMS is the availability of daily, cow-level data that are collected by AMS and associated technologies. As a result,
      • Tse C.
      • Barkema H.W.
      • Devries T.J.
      • Rushen J.
      • Pajor E.A.
      Effect of transitioning to automatic milking systems on producers' perceptions of farm management and cow health in the Canadian dairy industry.
      reported that, after transitioning to AMS, 66% of producers changed their health management strategy and 80% of producers found illness detection to be easier than before transitioning because of the AMS and associated health-monitoring software. On the other hand, some of the main barriers preventing adoption of technology by dairy producers are that technologies are not easy to use, they provide too much information without clear recommended action, and that their performance must be evaluated by independent research (
      • Russell R.A.
      • Bewley J.M.
      Characterization of Kentucky dairy producer decision-making behavior.
      ;
      • Borchers M.R.
      • Bewley J.M.
      An assessment of producer precision dairy farming technology use, prepurchase considerations, and usefulness.
      ).
      Thus, there is a need to transform behavior and production data into timely, useful, reliable, and actionable information for producers. Not only should these data be collected by validated technologies, they must be incorporated into validated models and algorithms by combining the knowledge of field experience and science. Producers must adapt their management skills to become more technology-based as they spend more time viewing and interpreting data. Furthermore, farmers, advisors, and support staff must learn to interpret information correctly, and with this information, they can implement the proper herd management and corrective action needed to achieve success with AMS.
      Therefore, this review summarizes research over the past 20 yr pertaining to the effects of health disorders on behavior and productivity of dairy cows, and the resultant use of those variables to help detect disorders. However, because such a large focus has been placed on using milk quality variables, such as electrical conductivity, SCC, and color, to detect mastitis in previous research (
      • Hogeveen H.
      • Kamphuis C.
      • Steeneveld W.
      • Mollenhorst H.
      Sensors and clinical mastitis—The quest for the perfect alert.
      ;
      • Rutten C.J.
      • Velthuis A.G.J.
      • Steeneveld W.
      • Hogeveen H.
      Invited review: sensors to support health management on dairy farms.
      ), this review will focus on using routinely collected behavior and production measures to detect locomotion and metabolic disorders such as lameness, hoof disorders, ketosis, subclinical ketosis (SCK), displaced abomasum (DA), metritis, and pneumonia, in addition to briefly discussing mastitis detection.
      Literature search criteria consisted of a web-based search through Web of Science, using the following search terms as topics: “automated milking” or “automatic milking” or “robotic milking,” and “dairy cow behavior” or “dairy cow production” or “dairy cow milk yield,” as well as a search regarding health management and illness detection. Inclusion criteria were that the paper must have been published in or after 2000 and must report on data collected routinely by AMS, such as milk yield, milk quality, BW, and cow activity and rumination behavior as measured by leg pedometers or neck collars.

      EFFECTS OF HEALTH DISORDERS ON BEHAVIOR AND PRODUCTION

      The negative effects of health disorders have been well documented for conventional herds, but less is known about these effects in AMS herds. The general outcomes associated with lameness and illness are likely similar in loose-housing systems, regardless of milking equipment; however, the individualized and voluntary nature of milking in AMS could intensify the effects of and responses to illness, given that cows are not manually brought to a milking parlor at set intervals. Therefore, we have comparatively summarized the consequences of lameness and illness in conventional and AMS herds to report overall trends, similarities, and differences.

      Lameness: Associations with Behavior and Production in AMS and Conventional Herds

      Table 1 shows recent findings regarding associations of lameness with behavior and productivity in conventional and AMS herds in the past 2 decades. Regarding milk yield, lameness in both conventional and AMS herds has obvious negative impacts, whether comparing lame and sound cows or looking at changes leading up to diagnosis. However, there is no clear effect on rumination time or activity (Table 1). Researchers have reported lower milk yield to be associated with lameness in conventional herds (4 to 10 kg/d lower than sound cows;
      • Van Hertem T.
      • Maltz E.
      • Antler A.
      • Romanini C.E.B.
      • Viazzi S.
      • Bahr C.
      • Schlageter-Tello A.
      • Lokhorst C.
      • Berckmans D.
      • Halachmi I.
      Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
      ) and AMS herds (1.6 kg/d lower than sound cows;
      • Bach A.
      • Dinarés M.
      • Devant M.
      • Carré X.
      Associations between lameness and production, feeding and milking attendance of Holstein cows milked with an automatic milking system.
      ;
      • King M.T.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • Devries T.J.
      Cow-level associations of lameness, behavior, and milk yield of cows milked in automated systems.
      ); leading up to lameness, milk yield of lame cows declined by 4 kg in total over 14 d (
      • Van Hertem T.
      • Maltz E.
      • Antler A.
      • Romanini C.E.B.
      • Viazzi S.
      • Bahr C.
      • Schlageter-Tello A.
      • Lokhorst C.
      • Berckmans D.
      • Halachmi I.
      Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
      ). Lame cows in a conventional herd had numerically (but not statistically significant) lower milk yield compared with healthy cows (
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health: the effect of post-calving health problems on rumination duration, activity and milk yield.
      ) and there was no association between milk yield and gait score in an AMS herd, when the majority of cows had locomotion score of 2 or 3 out of 5 (
      • Deming J.A.
      • Bergeron R.
      • Leslie K.E.
      • DeVries T.J.
      Associations of cow-level factors, frequency of feed delivery, and standing and lying behaviour of dairy cows milked in an automatic system.
      ).
      • Bicalho R.C.
      • Warnick L.D.
      • Guard C.L.
      Strategies to analyze milk losses caused by diseases with potential incidence throughout the lactation: A lameness example.
      conducted multiple analyses, using various study designs, to assess the impact of hoof horn lesions on milk yield in conventional herds. Because lame cows produced 3.2 kg/d more milk than control cows in the first 3 wk of lactation, those authors then controlled for that early-lactation milk yield and found that lame cows actually produced 1.0 kg/d less milk than control cows throughout lactation. The authors then matched 603 lame cows with 603 sound cows, again accounting for early-lactation milk yield, and found that lame cows produced 1.4 kg/d less milk than control cows (
      • Bicalho R.C.
      • Warnick L.D.
      • Guard C.L.
      Strategies to analyze milk losses caused by diseases with potential incidence throughout the lactation: A lameness example.
      ). Thus, milk yield may be greater in cows about to become lame (1.1 kg/d more milk before cows were diagnosed with lameness), but once diagnosed, their production drops to that of an average cow (
      • Green L.E.
      • Hedges V.J.
      • Schukken Y.H.
      • Blowey R.W.
      • Packington A.J.
      The impact of clinical lameness on the milk yield of dairy cows.
      ), and it is important to consider the previous milk yield and lactation potential of a cow when considering her current milk production.
      Table 1Associations of lameness with variables of interest in conventional and automatic milking system (AMS) herds
      TrialComparison(s)Change(s) associated with lamenessSample population
      Conventional herds
      • Almeida P.E.
      • Weber P.S.D.
      • Burton J.L.
      • Zanella A.J.
      Depressed DHEA and increased sickness response behaviors in lame dairy cows with inflammatory foot lesions.
      Lame vs. soundRumination time ↓16 cows, 1 farm
      • Bicalho R.C.
      • Warnick L.D.
      • Guard C.L.
      Strategies to analyze milk losses caused by diseases with potential incidence throughout the lactation: A lameness example.
      Lame vs. soundMilk yield ↓3,623 cows, 1 farm
      • Walker S.L.
      • Smith R.F.
      • Routly J.E.
      • Jones D.N.
      • Morris M.J.
      • Dobson H.
      Lameness, activity time-budgets, and estrus expression in dairy cattle.
      Lame vs. soundRumination time: no association Time spent walking ↓59 cows, 1 farm
      • Singh Y.
      • Lathwal S.S.
      • Chakravarty A.K.
      • Gupta A.K.
      • Mohanty T.K.
      • Raja T.V.
      • Dangi R.L.
      • Roy B.K.
      Effect of lameness (hoof disorders) on productivity of Karan Fries crossbred cows.
      Lame vs. soundMilk yield ↓163 cows, 1 farm
      • Van Hertem T.
      • Maltz E.
      • Antler A.
      • Romanini C.E.B.
      • Viazzi S.
      • Bahr C.
      • Schlageter-Tello A.
      • Lokhorst C.
      • Berckmans D.
      • Halachmi I.
      Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
      Lame vs. sound Within lame cows Relative to diagnosisMilk yield ↓ Activity ratio night:day ↑ Rumination time overnight ↓118 cows, 1 farm
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      Lame vs. sound and within lame relative to diagnosisRumination time ↓198 cows, 1 farm
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health: the effect of post-calving health problems on rumination duration, activity and milk yield.
      Lame vs. other vs. healthy by DIMRumination time ↓ Activity ↓ Milk yield: no association704 cows, 1 farm
      • Weigele H.C.
      • Gygax L.
      • Steiner A.
      • Wechsler B.
      • Burla J.B.
      Moderate lameness leads to marked behavioral changes in dairy cows.
      Moderately lame vs. soundActivity ↓ first hour after feed delivery or pushup Rumination time: no association389 cows, 17 farms
      AMS herds
      • Bach A.
      • Dinarés M.
      • Devant M.
      • Carré X.
      Associations between lameness and production, feeding and milking attendance of Holstein cows milked with an automatic milking system.
      Linear models with NRS
      NRS = numerical rating score.
      1 to 5
      Milk yield ↓ Total milking frequency and voluntary milkings ↓120 cows, 1 farm
      • Borderas T.F.
      • Fournier A.
      • Rushen J.
      • de Passillé A.M.B.
      Effect of lameness on dairy cows' visits to automatic milking systems.
      High vs. low frequency visitorsNRS ↓ for cows visiting more AMS frequently256 cows, 8 farms
      • Deming J.A.
      • Bergeron R.
      • Leslie K.E.
      • DeVries T.J.
      Associations of cow-level factors, frequency of feed delivery, and standing and lying behaviour of dairy cows milked in an automatic system.
      Linear models with NRS 1 to 5Milking frequency ↓ Milk yield: no association90 cows, 1 farm
      • Garcia E.
      • Klaas I.
      • Amigo J.M.
      • Bro R.
      • Enevoldsen C.
      Lameness detection challenges in automated milking systems addressed with partial least squares discriminant analysis.
      Lame vs. soundActivity in early morning/evening ↑ Activity unstable throughout the day Teat cup attachment speeds unstable88 cows, 1 farm
      • Miguel-Pacheco G.G.
      • Kaler J.
      • Remnant J.
      • Cheyne L.
      • Abbott C.
      • French A.P.
      • Pridmore T.P.
      • Huxley J.N.
      Behavioural changes in dairy cows with lameness in an automatic milking system.
      Lame vs. soundRumination time ↓ AMS visits ↓ overall and overnight Refusals and fetches: no significance150 cows, 1 farm
      • King M.T.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • Devries T.J.
      Cow-level associations of lameness, behavior, and milk yield of cows milked in automated systems.
      Lame vs. soundMilk yield, frequency, voluntary milkings and refusals (i.e., no milking permission) ↓ Activity ratio night:day ↑ Rumination time, activity, and night:day rumination ratio: no association1,184 cows, 41 farms
      1 NRS = numerical rating score.
      Some researchers have identified associations between lameness and rumination time in conventional and AMS herds. In some studies, lame cows spent less time ruminating than healthy animals in conventional systems (−10% or approximately 40–50 min/d;
      • Almeida P.E.
      • Weber P.S.D.
      • Burton J.L.
      • Zanella A.J.
      Depressed DHEA and increased sickness response behaviors in lame dairy cows with inflammatory foot lesions.
      ;
      • Van Hertem T.
      • Maltz E.
      • Antler A.
      • Romanini C.E.B.
      • Viazzi S.
      • Bahr C.
      • Schlageter-Tello A.
      • Lokhorst C.
      • Berckmans D.
      • Halachmi I.
      Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
      ;
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      ;
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health: the effect of post-calving health problems on rumination duration, activity and milk yield.
      ), whereas other researchers have observed no difference between lame and sound cows in conventional and AMS herds (
      • Walker S.L.
      • Smith R.F.
      • Routly J.E.
      • Jones D.N.
      • Morris M.J.
      • Dobson H.
      Lameness, activity time-budgets, and estrus expression in dairy cattle.
      ;
      • King M.T.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • Devries T.J.
      Cow-level associations of lameness, behavior, and milk yield of cows milked in automated systems.
      ). Before diagnosis of lameness or hoof disorders, researchers found that rumination time began to decline 3 to 5 d before diagnosis in a conventional herd by ∼50 min in total (n = 25;
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      ). In an AMS herd, rumination time declined throughout the 2-wk period before diagnosis by ∼40 min in total, controlling for DIM and parity (n = 15;
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      ); the latter group also found that the ratio of night:day rumination time increased leading up to diagnosis. Alternatively,
      • Van Hertem T.
      • Maltz E.
      • Antler A.
      • Romanini C.E.B.
      • Viazzi S.
      • Bahr C.
      • Schlageter-Tello A.
      • Lokhorst C.
      • Berckmans D.
      • Halachmi I.
      Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
      reported lame cows to have greater daytime rumination, lower nighttime rumination, and a lower ratio of night:day rumination compared with sound cows in a conventional herd, and although not analyzed specifically, rumination time of lame cows declined before diagnosis.
      With regards to activity, lame cows can be less active than healthy individuals. This has been demonstrated using data from neck collars to measure activity; differences have been seen in overall daily activity in conventional herds (−10%;
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health: the effect of post-calving health problems on rumination duration, activity and milk yield.
      ), as well as activity overnight in AMS (
      • Garcia E.
      • Klaas I.
      • Amigo J.M.
      • Bro R.
      • Enevoldsen C.
      Lameness detection challenges in automated milking systems addressed with partial least squares discriminant analysis.
      ;
      • King M.T.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • Devries T.J.
      Cow-level associations of lameness, behavior, and milk yield of cows milked in automated systems.
      ) and a conventional herd (
      • Van Hertem T.
      • Maltz E.
      • Antler A.
      • Romanini C.E.B.
      • Viazzi S.
      • Bahr C.
      • Schlageter-Tello A.
      • Lokhorst C.
      • Berckmans D.
      • Halachmi I.
      Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
      ). When assessed visually during specific observation periods, lame cows were observed walking less often than sound cows in a conventional system (
      • Walker S.L.
      • Smith R.F.
      • Routly J.E.
      • Jones D.N.
      • Morris M.J.
      • Dobson H.
      Lameness, activity time-budgets, and estrus expression in dairy cattle.
      ). In contrast,
      • King M.T.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • Devries T.J.
      Cow-level associations of lameness, behavior, and milk yield of cows milked in automated systems.
      saw no differences in daily activity between lame and sound individuals, and
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      did not observe changes in activity in the 2 wk before hoof disorders diagnoses.
      One might expect that overnight activity in AMS is related to milking activity and frequency. However, the increased nighttime activity for lame cows in AMS did not coincide with increased milking activity in 2 studies.
      • Miguel-Pacheco G.G.
      • Kaler J.
      • Remnant J.
      • Cheyne L.
      • Abbott C.
      • French A.P.
      • Pridmore T.P.
      • Huxley J.N.
      Behavioural changes in dairy cows with lameness in an automatic milking system.
      found that, compared with sound cows, lame cows not only visited the AMS less often overall 2.8 vs. 3.2 visits/d), they did so less between midnight and 0600 h. Similarly,
      • King M.T.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • Devries T.J.
      Cow-level associations of lameness, behavior, and milk yield of cows milked in automated systems.
      reported lame cows in AMS to have a lower milking frequency overall 0.3 fewer milkings/d) and as a nighttime:daytime ratio. In the 2 wk before diagnosis of hoof disorders in AMS, milking frequency declined by 0.05 milkings/d, a total reduction of approximately 0.6 daily milkings during this period (
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      ). Other researchers have also observed lower milking frequencies (overall and voluntary) for lame cows in AMS herds compared with sound cows (
      • Bach A.
      • Dinarés M.
      • Devant M.
      • Carré X.
      Associations between lameness and production, feeding and milking attendance of Holstein cows milked with an automatic milking system.
      ;
      • Borderas T.F.
      • Fournier A.
      • Rushen J.
      • de Passillé A.M.B.
      Effect of lameness on dairy cows' visits to automatic milking systems.
      ).
      • King M.T.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • Devries T.J.
      Cow-level associations of lameness, behavior, and milk yield of cows milked in automated systems.
      also reported 0.2 fewer refused milkings per day and 2.2× higher odds of fetched milkings for lame cows than for sound cows. Perhaps due to their smaller sample size for refusal and fetching observations,
      • Miguel-Pacheco G.G.
      • Kaler J.
      • Remnant J.
      • Cheyne L.
      • Abbott C.
      • French A.P.
      • Pridmore T.P.
      • Huxley J.N.
      Behavioural changes in dairy cows with lameness in an automatic milking system.
      did not observe any differences in frequencies of refusals or fetched milkings.
      For cows milked by an AMS, when accounting for DIM and parity, milk temperature began to increase starting 5 d before diagnosis of hoof disorders, whereas BW began to decrease starting 4 d before diagnosis (
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      ). New cases of lameness in AMS were generally associated with lower milk yield, rumination time, milk temperature, supplement intake, and milking and refusal frequencies compared with healthy cows; for lame cows, milk temperature was the only variable that deviated negatively from the baseline trajectory, accounting for parity and DIM (
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Behavior and productivity of cows milked in automated systems prior to diagnosis of health disorders in early lactation.
      ).
      Caveats to note regarding discrepancies within and between studies include those that controlled for DIM but did not necessary balance lame and sound groups (
      • Almeida P.E.
      • Weber P.S.D.
      • Burton J.L.
      • Zanella A.J.
      Depressed DHEA and increased sickness response behaviors in lame dairy cows with inflammatory foot lesions.
      ); those that excluded cows <40 DIM (
      • Van Hertem T.
      • Maltz E.
      • Antler A.
      • Romanini C.E.B.
      • Viazzi S.
      • Bahr C.
      • Schlageter-Tello A.
      • Lokhorst C.
      • Berckmans D.
      • Halachmi I.
      Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
      ), a period when illness is most likely; and those that excluded cows with moderate lameness (locomotion score of 2 out of 5) and parity >2 (
      • Garcia E.
      • Klaas I.
      • Amigo J.M.
      • Bro R.
      • Enevoldsen C.
      Lameness detection challenges in automated milking systems addressed with partial least squares discriminant analysis.
      ). Some researchers have analyzed cows with lameness and other health disorders (2 of 14 cases in
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      ), whereas others excluded cases of more than one health disorder (
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health: the effect of post-calving health problems on rumination duration, activity and milk yield.
      ); the first scenario is problematic if the other disorder affected variables of interest, and neither scenario allows for differentiation between types of illness. There is also variation between studies regarding the types of comparisons made, whether between lame and sound animals or within lame cows, as well as whether data were analyzed relative to the day of calving or to the day of diagnosis. Nonetheless, clear changes in behavior and productivity are associated with lameness that can be used to earlier detect, if not predict, lameness. Future studies should work to incorporate data representing the whole herd (i.e., fresh cows, all lactation groups, and other illnesses).

      Other Health Disorders: Associations with Behavior and Production in AMS and Conventional Herds

      Table 2 highlights the main changes in behavior and productivity associated with illness in conventional milking systems and AMS. General trends for activity, rumination time, and milk yield are consistent between studies, whether comparing sick and healthy animals, or looking within affected cows relative to diagnosis day.
      Table 2Associations of illness with variables of interest in conventional and automatic milking system (AMS) herds
      Freestall housing unless specified otherwise; MP = multiparous, PP = primiparous, DD = digestive disorders, SCK = subclinical ketosis, SCK+ = SCK plus another health disorder, SCH = subclinical hypocalcemia, RP = retained placenta, DA = displaced abomasum.
      TrialComparison(s)Change(s) associated with illnessSample population
      Conventional herds
      • Bareille N.
      • Beaudeau F.
      • Billon S.
      • Robert A.
      • Faverdin P.
      Effects of health disorders on feed intake and milk production in dairy cows.
      Within sick cows relative to diagnosisMilk yield ↓551 cows, 1 farm
      • Edwards J.L.
      • Tozer P.R.
      Using activity and milk yield as predictors of fresh cow disorders.
      Within sick cows relative to diagnosis Healthy vs. sick by DIMMilk yield ↓ Activity (steps/h) ↓1,445 cows, 3 farms
      • Huzzey J.M.
      • Veira D.M.
      • Weary D.M.
      • von Keyserlingk M.A.G.
      Prepartum behavior and dry matter intake identify dairy cows at risk for metritis.
      Metritis vs. healthy by DIMMilk yield ↓62 cows, 1 farm
      • DeVries T.J.
      • Beauchemin K.A.
      • Dohme F.
      • Schwartzkopf-Genswein K.S.
      Repeated ruminal acidosis challenges in lactating dairy cows at high and low risk for developing acidosis: Feeding, ruminating, and lying behavior.
      Relative to acidosis inductionRumination time ↓8 cows, 1 farm (tie-stall)
      • Goldhawk C.
      • Chapinal N.
      • Veira D.M.
      • Weary D.M.
      • von Keyserlingk M.A.G.
      Prepartum feeding behavior is an early indicator of subclinical ketosis.
      SCK vs. healthy by DIMMilk yield/daily change: no association62 cows, 1 farm
      • Lukas J.M.
      • Reneau J.K.
      • Wallace R.
      • Hawkins D.
      • Munoz-Zanzi C.
      A novel method of analyzing daily milk production and electrical conductivity to predict disease onset.
      Within sick cows relative to diagnosisMilk yield ↓ Milk conductivity ↑
      • Siivonen J.
      • Taponen S.
      • Hovinen M.
      • Pastell M.
      • Lensink B.J.
      • Pyörälä S.
      • Hänninen L.
      • Fogsgaard K.K.
      • Røntved C.M.
      • Sørensen P.
      • Herskin M.S.
      • Haley D.B.
      • Rushen J.
      • De Passillé A.M.
      Impact of acute clinical mastitis on cow behaviour.
      Relative to mastitis inductionDaily rumination time: no association 2-h rumination time intervals ↓ 4 and 8 h after Activity (steps/d) ↑6 cows, 1 farm (tie-stall)
      • Wittrock J.M.
      • Proudfoot K.L.
      • Weary D.M.
      • von Keyserlingk M.A.G.
      Short communication: Metritis affects milk production and cull rate of Holstein multiparous and primiparous dairy cows differently.
      Metritis vs. healthy by DIMMilk yield ↓ for MP cows, but not PP cows84 cows, 1 farm
      • Fogsgaard K.K.
      • Røntved C.M.
      • Sørensen P.
      • Herskin M.S.
      Sickness behavior in dairy cows during Escherichia coli mastitis.
      Relative to mastitis inductionMilk yield ↓ Rumination time ↓20 cows, 1 farm (tie-stall)
      • Jawor P.E.
      • Huzzey J.M.
      • LeBlanc S.J.
      • von Keyserlingk M.A.G.
      Associations of subclinical hypocalcemia at calving with milk yield, and feeding, drinking, and standing behaviors around parturition in Holstein cows.
      SCH vs. healthy by DIMMilk yield ↑ during wk 2–4 of lactation30 cows, 1 farm
      • Gáspárdy A.
      • Efrat G.
      • Bajcsy A.C.
      • Fekete S.G.
      Electronic monitoring of rumination activity as an indicator of health status and production traits in high-yielding dairy cows.
      Deviations relative to days surrounding diagnosis (d −3 to +3 vs. d −6 to −4 and d +4 to +6)Rumination time ↓ with SCK (PP and MP cows) BW ↓ with metritis (PP cows only) Rumination time ↓ with metritis (MP cows only)20 cows with SCK, 84 cows with metritis, 3 farms
      • Fogsgaard K.K.
      • Bennedsgaard T.W.
      • Herskin M.S.
      Behavioral changes in freestall-housed dairy cows with naturally occurring clinical mastitis.
      Mastitis vs. healthy after diagnosisActivity (steps/h) ↓60 cows, 1 farm
      • Liboreiro D.N.
      • Machado K.S.
      • Silva P.R.B.
      • Maturana M.M.
      • Nishimura T.K.
      • Brandão A.P.
      • Endres M.I.
      • Chebel R.C.
      Characterization of peripartum rumination and activity of cows diagnosed with metabolic and uterine diseases.
      RP, metritis, SCK vs. healthy by DIMRumination time and activity pre-calving: no association Rumination time and activity post-calving ↓296 cows, 1 farm
      • Kaufman E.I.
      • LeBlanc S.J.
      • McBride B.W.
      • Duffield T.F.
      • DeVries T.J.
      Association of rumination time with subclinical ketosis in transition dairy cows.
      SCK and SCK+ vs. healthy by DIMRumination time pre- and post-calving for SCK+ cows ↓202 MP cows, 4 farms
      • Schirmann K.
      • Weary D.M.
      • Heuwieser W.
      • Chapinal N.
      • Cerri R.L.A.
      • von Keyserlingk M.A.G.
      Short communication: Rumination and feeding behaviors differ between healthy and sick dairy cows during the transition period.
      SCK vs. healthy by DIMRumination time pre-calving ↓ Rumination time post-calving: no association80 cows, 1 farm
      • Stangaferro M.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part II. Mastitis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part III. Metritis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part I. Metabolic and digestive disorders.
      )
      Ketosis, mastitis, DA, metritis vs. healthy relative to day of diagnosis (from d −5 to +5)Rumination time ↓ for all Activity ↓ for all Milk yield ↓ all except metritis1,121 cows, 1 farm
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      Within sick cows relative to diagnosis Affected vs. healthyRumination time ↓198 cows, 1 farm
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health: the effect of post-calving health problems on rumination duration, activity and milk yield.
      Affected vs. healthy by DIM Within cow relative to diagnosisRumination time and activity ↓ No differences within cow704 cows, 1 farm
      AMS herds
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      Within sick cows relative to diagnosisMilk yield, frequency, rumination time, activity ↓57 cows, 1 farm
      • Veissier I.
      • Mialon M.M.
      • Sloth K.H.
      Short communication: Early modification of the circadian organization of cow activity in relation to disease or estrus.
      Within sick cows relative to diagnosisActivity patterns (circadian rhythm) varied 1–2 d before mastitis and lameness350 cows, 1 farm
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Behavior and productivity of cows milked in automated systems prior to diagnosis of health disorders in early lactation.
      Within sick cows relative to diagnosis Affected vs. healthyMilk yield, frequency, rumination time, activity ↓ Milk conductivity ↑ and temperature was variable605 cows, 9 farms
      1 Freestall housing unless specified otherwise; MP = multiparous, PP = primiparous, DD = digestive disorders, SCK = subclinical ketosis, SCK+ = SCK plus another health disorder, SCH = subclinical hypocalcemia, RP = retained placenta, DA = displaced abomasum.
      In conventional herds,
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health: the effect of post-calving health problems on rumination duration, activity and milk yield.
      found reductions in milk yield, rumination time, and activity beginning 5 d before diagnosis of several disorders. Milk yield declined 1 to 2 d before diagnosis of milk fever, udder edema, ketosis, and systemic mastitis; 4 d before puerperal metritis, digestive disorders, hock lesions, and recurrence of hock lesions or digestive disorders; and 5 d before foot lesions, chronic metritis, recurrence of ketosis, mastitis, or diarrhea (
      • Bareille N.
      • Beaudeau F.
      • Billon S.
      • Robert A.
      • Faverdin P.
      Effects of health disorders on feed intake and milk production in dairy cows.
      ). In another study, milk yield declined 1 to 2 d before milk fever, edema, mastitis, and hoof treatments; 4 to 9 d before digestive disorders, pneumonia, lameness, and abscess; and ≥10 d before going off feed or developing ketosis or metritis (
      • Lukas J.M.
      • Reneau J.K.
      • Wallace R.
      • Hawkins D.
      • Munoz-Zanzi C.
      A novel method of analyzing daily milk production and electrical conductivity to predict disease onset.
      ). In the latter study, milk conductivity also increased 1 to 3 d before milk fever, DA, edema, mastitis, lameness, and RP; and 6 to 9 d before pneumonia, off feed, and ketosis.
      • Soriani N.
      • Trevisi E.
      • Calamari L.
      Relationships between rumination time, metabolic conditions, and health status in dairy cows during the transition period.
      grouped cows by their precalving rumination time; compared with cows with high precalving rumination time, cows with lower rumination time before calving maintained that lower rumination time after calving and had a higher incidence of clinical disease (retained placenta, mastitis, ketosis, metritis, DA, and lameness).
      Before metabolic disorders specifically, activity (steps/d) began to decline 9 d before ketosis, DA, and digestive disorders; and milk yield began to drop 5 d before diagnosis (
      • Edwards J.L.
      • Tozer P.R.
      Using activity and milk yield as predictors of fresh cow disorders.
      ).
      • Stangaferro M.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part II. Mastitis.
      reported reductions in rumination time and activity occurring 1 to 5 d before diagnosis of metabolic disorders. After inducing subacute ruminal acidosis,
      • DeVries T.J.
      • Beauchemin K.A.
      • Dohme F.
      • Schwartzkopf-Genswein K.S.
      Repeated ruminal acidosis challenges in lactating dairy cows at high and low risk for developing acidosis: Feeding, ruminating, and lying behavior.
      found that rumination time declined by >1.5 h/d. Although
      • Goldhawk C.
      • Chapinal N.
      • Veira D.M.
      • Weary D.M.
      • von Keyserlingk M.A.G.
      Prepartum feeding behavior is an early indicator of subclinical ketosis.
      did not observe differences in milk yield between healthy cows and those with SCK (serum BHB concentrations ≥1.0 mmol/L during wk 1 postcalving),
      • Gáspárdy A.
      • Efrat G.
      • Bajcsy A.C.
      • Fekete S.G.
      Electronic monitoring of rumination activity as an indicator of health status and production traits in high-yielding dairy cows.
      reported a negative deviation in milk yield and rumination time with onset of SCK (unknown diagnosis criteria). Associations of SCK with rumination time and activity before and after calving vary by study.
      • Liboreiro D.N.
      • Machado K.S.
      • Silva P.R.B.
      • Maturana M.M.
      • Nishimura T.K.
      • Brandão A.P.
      • Endres M.I.
      • Chebel R.C.
      Characterization of peripartum rumination and activity of cows diagnosed with metabolic and uterine diseases.
      found no rumination or activity differences precalving (BHB >1.0 mmol/L during wk 1–3 postcalving), but those authors found postcalving reductions in rumination time up to 80 min/d and in activity up to 110 daily units. In contrast,
      • Schirmann K.
      • Weary D.M.
      • Heuwieser W.
      • Chapinal N.
      • Cerri R.L.A.
      • von Keyserlingk M.A.G.
      Short communication: Rumination and feeding behaviors differ between healthy and sick dairy cows during the transition period.
      saw no differences postcalving and only observed 1 h/d lower rumination times before calving for cows that later had SCK after calving (BHB ≥1.2 mmol/L during the first 14 d). On the other hand,
      • Kaufman E.I.
      • LeBlanc S.J.
      • McBride B.W.
      • Duffield T.F.
      • DeVries T.J.
      Association of rumination time with subclinical ketosis in transition dairy cows.
      reported that multiparous cows with SCK (BHB >1.2 mmol/L during wk 1–4 postcalving) and another health disorder spent 50 and 90 min/d less time ruminating, respectively, pre- and postcalving compared with healthy multiparous cows.
      Numerous researchers have documented changes associated with metritis in conventional herds (
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      ), with varying degrees of response in milk yield, BW, rumination time, and activity based on the severity of metritis (
      • Huzzey J.M.
      • Veira D.M.
      • Weary D.M.
      • von Keyserlingk M.A.G.
      Prepartum behavior and dry matter intake identify dairy cows at risk for metritis.
      ;
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part I. Metabolic and digestive disorders.
      ); results also differed by parity, but overall the trends were similar looking at days preceding and following illness (
      • Gáspárdy A.
      • Efrat G.
      • Bajcsy A.C.
      • Fekete S.G.
      Electronic monitoring of rumination activity as an indicator of health status and production traits in high-yielding dairy cows.
      ). Researchers studying naturally occurring mastitis (
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part III. Metritis.
      ;
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      ) and those cases induced using an endotoxin (
      • Siivonen J.
      • Taponen S.
      • Hovinen M.
      • Pastell M.
      • Lensink B.J.
      • Pyörälä S.
      • Hänninen L.
      • Fogsgaard K.K.
      • Røntved C.M.
      • Sørensen P.
      • Herskin M.S.
      • Haley D.B.
      • Rushen J.
      • De Passillé A.M.
      Impact of acute clinical mastitis on cow behaviour.
      ;
      • Fogsgaard K.K.
      • Røntved C.M.
      • Sørensen P.
      • Herskin M.S.
      Sickness behavior in dairy cows during Escherichia coli mastitis.
      ) have documented reductions in rumination time, milk yield, and activity leading up to the day of diagnosis or induction, ranging from 2 to 7 h/d rumination time (
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      ).
      To summarize the changes in behavior and productivity with illness in AMS herds, we have included the only 3 articles we are aware of, which are consistent with reports in conventional herds.
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      examined numerous variables from 56 cases of illness in 1 herd relative to the day of diagnosis (d 0) for each illness separately. Always controlling for DIM and parity in their models, they assessed deviations relative to the baseline trajectory of cows affected with a specific disorder. Deviations in milk yield began 4 to 5 d before DA, pneumonia, and SCK diagnoses by 1.2 to 4.4 kg/d each day until d −1, with a total reduction of 4.8 to 13.2 kg/d. Rumination time began to decline 5 to 8 d before DA, pneumonia, and SCK diagnoses by 25 to 50 min/d each d until d −1, with a total reduction of 2 to 5 h/d. Similarly, activity reductions began 4 to 7 d before diagnoses of DA, pneumonia, SCK, and metritis by 20 to 40% total, and BW reductions began 4 to 6 d before pneumonia, SCK, hoof disorders, and metritis diagnoses by 10 to 14 kg/d each day. Only before DA, in that study, did milk temperature deviate 6 d before diagnosis (
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      ). The key finding in the
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      study was the consistent response of rumination time deviating at least 1 d before milk yield, on average, but that study also observed large variation between cows. In their follow-up study, with 605 cows in 9 herds, deviations were generally consistent (
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Behavior and productivity of cows milked in automated systems prior to diagnosis of health disorders in early lactation.
      ). Comparisons were also made with a healthy group of cows given a mock day of diagnosis, from which sick cows often differed even 2 wk before diagnosis (
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Behavior and productivity of cows milked in automated systems prior to diagnosis of health disorders in early lactation.
      ). Additional trends observed in that study included deviations from baseline milking frequency and milk conductivity occurring 8 and 12 d before diagnosis of mastitis, respectively. A contradictory result was that only the milk temperature of lame cows deviated from baseline and not cows with DA. The overall message from that work was that acute health disorders (i.e., DA and mastitis) were associated with deviations from those cows' baseline AMS data, whereas more chronic disorders (i.e., SCK and lameness) were associated with significant, but subtle, longer-term changes in milk production and behavior. Finally,
      • Veissier I.
      • Mialon M.M.
      • Sloth K.H.
      Short communication: Early modification of the circadian organization of cow activity in relation to disease or estrus.
      noted that the circadian activity of cows varied 1 to 2 d before diagnosis of lameness and mastitis.
      It would be interesting to compare behavioral responses to health disorders in both conventional and AMS herds. With the current data available, however, we cannot determine whether responses to lameness and illness are augmented in AMS. Furthermore, variation may exist between housing systems (bedded pack vs. freestall) and traffic type (forced vs. free). Therefore, more research is needed to collect information on large samples of cows from many farms of diverse barn designs and management practices.

      HEALTH MANAGEMENT SOFTWARE AND DETECTION MODELS

      A key feature of AMS is data collection and subsequent generation of management reports and alert lists. Data can be used to create management reports and task lists, as well as attention lists of cows with potential health problems. These reports can overwhelm producers with excessive alerts (false positives), while not necessarily being sensitive enough to pick up chronic disorders (false negatives). Thus, these data must be transformed into useful, reliable information for producers, using field experience and science-based recommendations. Despite some inconsistencies in the literature, the potential certainly exists to combine the use of several behavior and production variables to create highly sensitive and specific health detection models. The goal should be to achieve 99% specificity and 80% sensitivity, as recommended by the International Organization for Standardization, to evaluate model performance for automated detection of abnormal milk (i.e.,
      • Rutten C.J.
      • Velthuis A.G.J.
      • Steeneveld W.
      • Hogeveen H.
      Invited review: sensors to support health management on dairy farms.
      ). Additionally, illness detection software should include adjustable settings to personalize specificity and sensitivity based on each farmer's management strategy, such as how willing they are to take risks weighted against the time needed to visually assess flagged cows that may not actually be sick.
      Alerts created by AMS manufacturers are currently available on-farm but do not always incorporate validated models and algorithms using data from validated technologies. These reports include a mastitis detection index and a milking performance index (teat attachment speed) for herds with the DeLaval system (DelPro; DeLaval International AB, Tumba, Sweden). Herd Navigator (DeLaval International AB) measures lactate dehydrogenase for mastitis detection, milk BHB for ketosis detection, and milk progesterone for heat detection and pregnancy. Lely's T4C software (Time-for-Cows, Lely Industries N.V., Maassluis, the Netherlands) has milk, estrus, and rumination attention lists, which flag cows with deviations in milk yield, cell count, conductivity, and temperature, BW, activity, and rumination time. Within the last few years, Lely has released a sick cow report, which uses a combination of these variables. Although these metrics and commercial alerts have great potential to aid in illness detection, they have not been fully validated (or this has not been done so transparently); that is, peer-reviewed and publicly available scientific validation is not available. Nonetheless, many of these alerts are already in use in the field.
      The same variables used in commercial health alerts have been incorporated into various detection models created and validated by researchers.
      • Stangaferro M.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part II. Mastitis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part III. Metritis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part I. Metabolic and digestive disorders.
      ) assessed the validity of a health index score, composed of rumination and activity data during a 5-d period before diagnosis, in a conventional herd. The index achieved a sensitivity of 98% for DA, 91% for ketosis, 89% for indigestion, and 93% for those 3 disorders combined (
      • Stangaferro M.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part II. Mastitis.
      ). Sensitivity for all cases of clinical mastitis and metritis was 58 and 55%, respectively; although this was similar for cases of only mastitis (55%) or metritis (53%), sensitivity was greater for cows with mastitis and another health disorder (89%), and for cows with metritis and another disorder (78%;
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part III. Metritis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part I. Metabolic and digestive disorders.
      ). The index was most sensitive for cases of Escherichia coli mastitis (81%) and less so for gram-positive cases of mastitis (45 to 48%) and less severe cases of metritis (
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part III. Metritis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part I. Metabolic and digestive disorders.
      ). Considering all health disorders studied, those researchers reported an overall sensitivity of 59% and specificity of 98% for the health index score.
      • Steensels M.
      • Antler A.
      • Bahr C.
      • Berckmans D.
      • Maltz E.
      • Halachmi I.
      A decision-tree model to detect post-calving diseases based on rumination, activity, milk yield, BW and voluntary visits to the milking robot.
      used a decision tree model to detect ketosis and metritis in AMS-milked cows. With a specificity of 87% and sensitivity of 69%, their model considered rumination time and activity (measured by neck collar monitors), as well as milk yield, milk slope, and current BW relative to BW at calving (measured by AMS) to predict the probability of illness for each animal. It is unclear, however, which cows were considered healthy in that study, or whether cows with other disorders were excluded from the analysis. Those authors also evaluated a ketosis prediction model on 4 different farms (
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health; design and validation of a model to detect ketosis.
      ), finding that its sensitivity (78–90%) and specificity (71–74%) were improved when running model training and validation within the same herd. However,
      • Steensels M.
      • Maltz E.
      • Bahr C.
      • Berckmans D.
      • Antler A.
      • Halachmi I.
      Towards practical application of sensors for monitoring animal health; design and validation of a model to detect ketosis.
      and
      • Stangaferro M.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part II. Mastitis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part III. Metritis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part I. Metabolic and digestive disorders.
      ) excluded cows with multiple health disorders. Therefore, limitations remain with current prediction models regarding how to deal with cases of more than one illness and how to detect one illness without excluding the others from analyses.
      Regarding lameness detection,
      • Garcia E.
      • Klaas I.
      • Amigo J.M.
      • Bro R.
      • Enevoldsen C.
      Lameness detection challenges in automated milking systems addressed with partial least squares discriminant analysis.
      created models using activity (measured by neck collar monitors), milk flow, and teat cup attachment data from 1 wk before and after lameness diagnosis in an AMS herd. Separate models were created for first- and second-lactation cows, which had reported specificities of 77 and 83%, respectively, and both had a sensitivity of 79%. Using a model combining several activity variables (i.e., number of steps negatively associated with lameness, and lying behavior data with various associations, measured by leg band accelerometers), milk yield per day (≥3 negative outliers/week), and concentrate leftovers per day (positively associated with lameness),
      • de Mol R.M.
      • André G.
      • Bleumer E.J.B.
      • van der Werf J.T.N.
      • de Haas Y.
      • van Reenen C.G.
      Applicability of day-to-day variation in behavior for the automated detection of lameness in dairy cows.
      created a model with high sensitivity (85.5%) and specificity (88.8%) for identifying lameness on the same day as diagnosis in an AMS herd. Those 2 studies, however, did not include cows with an uneven gait that did not obviously favor one limb, which made up almost 50% of their observations; the difference between those and lame cows can be subtle, and those authors recognized this data exclusion as a large limitation of their study.
      • Van Hertem T.
      • Maltz E.
      • Antler A.
      • Romanini C.E.B.
      • Viazzi S.
      • Bahr C.
      • Schlageter-Tello A.
      • Lokhorst C.
      • Berckmans D.
      • Halachmi I.
      Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
      also created a lameness detection model, which used milk yield, activity, and rumination data from 4 to 7 d before diagnosis in a parlor-milked herd. Those authors reported a sensitivity of 89% and specificity of 86%, finding that lame cows produced less milk, were proportionately less active during the day, and spent less time ruminating at night compared with sound cows. Other researchers have incorporated similar variables into lameness detection models, such as BW, activity, milk order, walking speed, standing bouts, lying time, and limb weight ratio during milking (
      • Kamphuis C.
      • Frank E.
      • Burke J.K.
      • Verkerk G.A.
      • Jago J.G.
      Applying additive logistic regression to data derived from sensors monitoring behavioral and physiological characteristics of dairy cows to detect lameness.
      ;
      • Beer G.
      • Alsaaod M.
      • Starke A.
      • Schuepbach-Regula G.
      • Mülller H.
      • Kohler P.
      • Steiner A.
      Use of extended characteristics of locomotion and feeding behavior for automated identification of lame dairy cows.
      ;
      • Nechanitzky K.
      • Starke A.
      • Vidondo B.
      • Müller H.
      • Reckardt M.
      • Friedli K.
      • Steiner A.
      Analysis of behavioral changes in dairy cows associated with claw horn lesions.
      ); however, those researchers excluded cows affected by health disorders other than lameness, which would reduce their false positive rates, and, in 2 studies, fresh cows <30 to 40 DIM (when many other health disorders arise) were excluded. Using cow activity and wavelet analysis,
      • Miekley B.
      • Traulsen I.
      • Krieter J.
      Detection of mastitis and lameness in dairy cows using wavelet analysis.
      created lameness detection models with 40 to 64% sensitivity, 72 to 85% specificity, and high error rates (>90%).
      For credibility, accuracy, and repeatability, science-based recommendations are needed to generate health attention reports and to select settings for on-farm application, such as the recent development (
      • Kamphuis C.
      • Dela Rue B.
      • Mein G.
      • Jago J.
      Development of protocols to evaluate in-line mastitis-detection systems.
      ) and field validation (
      • Kamphuis C.
      • Dela Rue B.T.
      • Eastwood C.R.
      Field validation of protocols developed to evaluate in-line mastitis detection systems.
      ) of an inline detection system for mastitis and high SCC; those researchers also consulted industry representatives for feedback to improve the application of those models. To briefly touch on mastitis detection in conventional milking systems,
      • Miekley B.
      • Traulsen I.
      • Krieter J.
      Mastitis detection in dairy cows: The application of support vector machines.
      achieved 85% sensitivity and 72 to 83% specificity when detecting mastitis using deviations from 5-d moving averages and standard deviations of milk yield and conductivity. In similar studies,
      • Miekley B.
      • Traulsen I.
      • Krieter J.
      Detection of mastitis and lameness in dairy cows using wavelet analysis.
      ,
      • Miekley B.
      • Traulsen I.
      • Krieter J.
      Principal component analysis for the early detection of mastitis and lameness in dairy cows.
      ) used milk yield and conductivity to detect mastitis using principal components and wavelet analyses and concluded that these methods were not ready for on-farm application. In AMS herds,
      • Sørensen L.P.
      • Bjerring M.
      • Løvendahl P.
      Monitoring individual cow udder health in automated milking systems using online somatic cell counts.
      evaluated online SCC data to identify elevated risk of mastitis in AMS. By modifying 2 reports available with that software, they tested various validation scenarios; based on new alerts, those models achieved sensitivities of 28 to 43% and specificities >99%, but they achieved much higher sensitivities when validating based on cows that appeared on the intramammary infection report (55–89%). Those authors are currently working toward the highly desired integration of information to help with decision-making and monitoring.

      FUTURE DIRECTIONS

      Remaining gaps in knowledge are how to best identify cows with lameness or other health disorders in AMS herds using a combination of rumination, activity, and milk data, especially for early lactation animals. Different variations (e.g., standard deviation, relative change) and weightings of each variable may also prove useful. Such a multivariate approach may enhance the performance of illness detection models (
      • Miekley B.
      • Traulsen I.
      • Krieter J.
      Detection of mastitis and lameness in dairy cows using wavelet analysis.
      ,
      • Miekley B.
      • Traulsen I.
      • Krieter J.
      Principal component analysis for the early detection of mastitis and lameness in dairy cows.
      ). Not only may health alerts notify producers of potential issues, but those alerts could indicate the probability and type of problem as well. This could even be a general category of illness; that is, metabolic versus infectious, internally versus externally caused, or acute versus state-like.
      Of the current studies that have worked toward disease detection, many data have been excluded or separated for isolated models (i.e., fresh cows, multiple lactation groups, and cows with multiple illnesses or moderate cases). Thus, future studies should incorporate data from the entire lactating herd and should consider, and account for, parity and stage of lactation or DIM of each animal, regardless of whether they retain those variables in final algorithms. Researchers have previously controlled for parity (
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      ) or even created different models for parity groups, finding activity data to be more informative for primiparous cows and less informative for second-lactation cows (
      • Garcia E.
      • Klaas I.
      • Amigo J.M.
      • Bro R.
      • Enevoldsen C.
      Lameness detection challenges in automated milking systems addressed with partial least squares discriminant analysis.
      ).
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      found differences in slope estimates when accounting for DIM, whereas
      • Garcia E.
      • Klaas I.
      • Amigo J.M.
      • Bro R.
      • Enevoldsen C.
      Lameness detection challenges in automated milking systems addressed with partial least squares discriminant analysis.
      saw no improvements in their models when they included DIM.
      Researchers have examined, and should continue to investigate, deviations within sick cows as an average (
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Behavior and productivity of cows milked in automated systems prior to diagnosis of health disorders in early lactation.
      ) or individually (
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      ). Potentially most importantly, cows with illness should be compared with a group of healthy contemporaries (
      • Stangaferro M.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part II. Mastitis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part III. Metritis.
      ,
      • Stangaferro M.L.
      • Wijma R.
      • Quinteros C.
      • Medrano M.
      • Masello M.
      • Giordano J.
      Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part I. Metabolic and digestive disorders.
      ;
      • Paudyal S.
      • Maunsell F.P.
      • Richeson J.T.
      • Risco C.A.
      • Donovan D.A.
      • Pinedo P.J.
      Rumination time and monitoring of health disorders during early lactation.
      ;
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Behavior and productivity of cows milked in automated systems prior to diagnosis of health disorders in early lactation.
      ), which could refine the ability of detection models to identify subtle deviations in early lactation.
      Finally, attention should be given to creating separate models for dry cows and young stock. It is likely that, as in lactating cows, the activity or behavior of those animals will respond before production variables begin to decline (
      • Edwards J.L.
      • Tozer P.R.
      Using activity and milk yield as predictors of fresh cow disorders.
      ;
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
      ,
      • King M.T.M.
      • Dancy K.M.
      • LeBlanc S.J.
      • Pajor E.A.
      • DeVries T.J.
      Behavior and productivity of cows milked in automated systems prior to diagnosis of health disorders in early lactation.
      ), including growth, body condition, or allocation of resources to a developing fetus. We encourage manufacturers and software makers to make (at least some of) their algorithm components publicly available to aid independent researchers in understanding and improving upon those methods.

      CONCLUSIONS

      This review summarizes 2 decades of research regarding the effects of health disorders on cow behavior and production, and the resultant use of those variables to help detect illness. Compared with behaviors in healthy cows, consistent trends are observed in activity, rumination time, and milk yield leading up to diagnosis of health disorders, except for lameness, which may not necessarily alter rumination time or activity. To advance this technology, it is necessary for commercial software to incorporate validated models and algorithms for generating health alerts on-farm. Future studies should (1) incorporate the entire lactating herd while accounting for stage of lactation and parity of each animal; (2) evaluate the deviations that cows exhibit from their own baseline trajectories and relative to healthy contemporaries; (3) combine the use of several variables into health alerts; and (4) differentiate the probable type of health disorder.

      ACKNOWLEDGMENTS

      Meagan King was financially supported in part by the Canada First Research Excellence Fund (Ottawa, ON, Canada) and by a contribution from the Dairy Research Cluster II Initiative, funded by the Dairy Farmers of Canada (Ottawa, ON, Canada), Agriculture and Agri-Food Canada (Ottawa, ON, Canada), the Canadian Dairy Network (Guelph, ON, Canada), and the Canadian Dairy Commission (Ottawa, ON, Canada).

      REFERENCES

        • Almeida P.E.
        • Weber P.S.D.
        • Burton J.L.
        • Zanella A.J.
        Depressed DHEA and increased sickness response behaviors in lame dairy cows with inflammatory foot lesions.
        Domest. Anim. Endocrinol. 2008; 34 (17229542): 89-99
        • Bach A.
        • Dinarés M.
        • Devant M.
        • Carré X.
        Associations between lameness and production, feeding and milking attendance of Holstein cows milked with an automatic milking system.
        J. Dairy Res. 2007; 74 (16978436): 40-46
        • Bareille N.
        • Beaudeau F.
        • Billon S.
        • Robert A.
        • Faverdin P.
        Effects of health disorders on feed intake and milk production in dairy cows.
        Livest. Prod. Sci. 2003; 83: 53-62
        • Barkema H.W.
        • von Keyserlingk M.A.G.
        • Kastelic J.P.
        • Lam T.J.G.M.
        • Luby C.
        • Roy J.-P.
        • LeBlanc S.J.
        • Keefe G.P.
        • Kelton D.F.
        Invited review: Changes in the dairy industry affecting dairy cattle health and welfare.
        J. Dairy Sci. 2015; 98 (26342982): 7426-7445
        • Beer G.
        • Alsaaod M.
        • Starke A.
        • Schuepbach-Regula G.
        • Mülller H.
        • Kohler P.
        • Steiner A.
        Use of extended characteristics of locomotion and feeding behavior for automated identification of lame dairy cows.
        PLoS One. 2016; 11 (27187073): e0155796
        • Bicalho R.C.
        • Warnick L.D.
        • Guard C.L.
        Strategies to analyze milk losses caused by diseases with potential incidence throughout the lactation: A lameness example.
        J. Dairy Sci. 2008; 91 (18565924): 2653-2661
        • Borchers M.R.
        • Bewley J.M.
        An assessment of producer precision dairy farming technology use, prepurchase considerations, and usefulness.
        J. Dairy Sci. 2015; 98 (25892693): 4198-4205
        • Borderas T.F.
        • Fournier A.
        • Rushen J.
        • de Passillé A.M.B.
        Effect of lameness on dairy cows' visits to automatic milking systems.
        Can. J. Anim. Sci. 2008; 88: 1-8
        • de Mol R.M.
        • André G.
        • Bleumer E.J.B.
        • van der Werf J.T.N.
        • de Haas Y.
        • van Reenen C.G.
        Applicability of day-to-day variation in behavior for the automated detection of lameness in dairy cows.
        J. Dairy Sci. 2013; 96 (23548300): 3703-3712
        • Deming J.A.
        • Bergeron R.
        • Leslie K.E.
        • DeVries T.J.
        Associations of cow-level factors, frequency of feed delivery, and standing and lying behaviour of dairy cows milked in an automatic system.
        Can. J. Anim. Sci. 2013; 93: 427-433
        • DeVries T.J.
        • Beauchemin K.A.
        • Dohme F.
        • Schwartzkopf-Genswein K.S.
        Repeated ruminal acidosis challenges in lactating dairy cows at high and low risk for developing acidosis: Feeding, ruminating, and lying behavior.
        J. Dairy Sci. 2009; 92 (19762825): 5067-5078
        • Edwards J.L.
        • Tozer P.R.
        Using activity and milk yield as predictors of fresh cow disorders.
        J. Dairy Sci. 2004; 87 (14762096): 524-531
        • Fogsgaard K.K.
        • Bennedsgaard T.W.
        • Herskin M.S.
        Behavioral changes in freestall-housed dairy cows with naturally occurring clinical mastitis.
        J. Dairy Sci. 2015; 98 (25547306): 1730-1738
        • Fogsgaard K.K.
        • Røntved C.M.
        • Sørensen P.
        • Herskin M.S.
        Sickness behavior in dairy cows during Escherichia coli mastitis.
        J. Dairy Sci. 2012; 95 (22281328): 630-638
        • Garcia E.
        • Klaas I.
        • Amigo J.M.
        • Bro R.
        • Enevoldsen C.
        Lameness detection challenges in automated milking systems addressed with partial least squares discriminant analysis.
        J. Dairy Sci. 2014; 97 (25282423): 7476-7486
        • Gáspárdy A.
        • Efrat G.
        • Bajcsy A.C.
        • Fekete S.G.
        Electronic monitoring of rumination activity as an indicator of health status and production traits in high-yielding dairy cows.
        Acta Vet. Hung. 2014; 62 (25410387): 452-462
        • Goldhawk C.
        • Chapinal N.
        • Veira D.M.
        • Weary D.M.
        • von Keyserlingk M.A.G.
        Prepartum feeding behavior is an early indicator of subclinical ketosis.
        J. Dairy Sci. 2009; 92 (19762814): 4971-4977
        • Green L.E.
        • Hedges V.J.
        • Schukken Y.H.
        • Blowey R.W.
        • Packington A.J.
        The impact of clinical lameness on the milk yield of dairy cows.
        J. Dairy Sci. 2002; 85 (12362457): 2250-2256
        • Hogeveen H.
        • Kamphuis C.
        • Steeneveld W.
        • Mollenhorst H.
        Sensors and clinical mastitis—The quest for the perfect alert.
        Sensors (Basel). 2010; 10 (22163637): 7991-8009
        • Huzzey J.M.
        • Veira D.M.
        • Weary D.M.
        • von Keyserlingk M.A.G.
        Prepartum behavior and dry matter intake identify dairy cows at risk for metritis.
        J. Dairy Sci. 2007; 90 (17582105): 3220-3233
        • Jacobs J.A.
        • Siegford J.M.
        Invited review: The impact of automatic milking systems on dairy cow management, behavior, health, and welfare.
        J. Dairy Sci. 2012; 95 (22541453): 2227-2247
        • Jawor P.E.
        • Huzzey J.M.
        • LeBlanc S.J.
        • von Keyserlingk M.A.G.
        Associations of subclinical hypocalcemia at calving with milk yield, and feeding, drinking, and standing behaviors around parturition in Holstein cows.
        J. Dairy Sci. 2012; 95 (22365207): 1240-1248
        • Kamphuis C.
        • Dela Rue B.
        • Mein G.
        • Jago J.
        Development of protocols to evaluate in-line mastitis-detection systems.
        J. Dairy Sci. 2013; 96 (23548290): 4047-4058
        • Kamphuis C.
        • Dela Rue B.T.
        • Eastwood C.R.
        Field validation of protocols developed to evaluate in-line mastitis detection systems.
        J. Dairy Sci. 2016; 99 (26686708): 1619-1631
        • Kamphuis C.
        • Frank E.
        • Burke J.K.
        • Verkerk G.A.
        • Jago J.G.
        Applying additive logistic regression to data derived from sensors monitoring behavioral and physiological characteristics of dairy cows to detect lameness.
        J. Dairy Sci. 2013; 96 (24011945): 7043-7053
        • Kaufman E.I.
        • LeBlanc S.J.
        • McBride B.W.
        • Duffield T.F.
        • DeVries T.J.
        Association of rumination time with subclinical ketosis in transition dairy cows.
        J. Dairy Sci. 2016; 99 (27132097): 5604-5618
        • King M.T.M.
        • Dancy K.M.
        • LeBlanc S.J.
        • Pajor E.A.
        • DeVries T.J.
        Deviations in behavior and productivity data before diagnosis of health disorders in cows milked with an automated system.
        J. Dairy Sci. 2017; 100 (28755933): 8358-8371
        • King M.T.M.
        • Dancy K.M.
        • LeBlanc S.J.
        • Pajor E.A.
        • DeVries T.J.
        Behavior and productivity of cows milked in automated systems prior to diagnosis of health disorders in early lactation.
        J. Dairy Sci. 2018; 101: 4343-4356
        • King M.T.M.
        • LeBlanc S.J.
        • Pajor E.A.
        • Devries T.J.
        Cow-level associations of lameness, behavior, and milk yield of cows milked in automated systems.
        J. Dairy Sci. 2017; 100 (28434734): 4818-4828
        • Liboreiro D.N.
        • Machado K.S.
        • Silva P.R.B.
        • Maturana M.M.
        • Nishimura T.K.
        • Brandão A.P.
        • Endres M.I.
        • Chebel R.C.
        Characterization of peripartum rumination and activity of cows diagnosed with metabolic and uterine diseases.
        J. Dairy Sci. 2015; 98 (26233440): 6812-6827
        • Lukas J.M.
        • Reneau J.K.
        • Wallace R.
        • Hawkins D.
        • Munoz-Zanzi C.
        A novel method of analyzing daily milk production and electrical conductivity to predict disease onset.
        J. Dairy Sci. 2009; 92 (19923600): 5964-5976
        • Miekley B.
        • Traulsen I.
        • Krieter J.
        Detection of mastitis and lameness in dairy cows using wavelet analysis.
        Livest. Sci. 2012; 148: 227-236
        • Miekley B.
        • Traulsen I.
        • Krieter J.
        Mastitis detection in dairy cows: The application of support vector machines.
        J. Agric. Sci. 2013; 151: 889-897
        • Miekley B.
        • Traulsen I.
        • Krieter J.
        Principal component analysis for the early detection of mastitis and lameness in dairy cows.
        J. Dairy Res. 2013; 80 (23823869): 335-343
        • Miguel-Pacheco G.G.
        • Kaler J.
        • Remnant J.
        • Cheyne L.
        • Abbott C.
        • French A.P.
        • Pridmore T.P.
        • Huxley J.N.
        Behavioural changes in dairy cows with lameness in an automatic milking system.
        Appl. Anim. Behav. Sci. 2014; 150: 1-8
        • Nechanitzky K.
        • Starke A.
        • Vidondo B.
        • Müller H.
        • Reckardt M.
        • Friedli K.
        • Steiner A.
        Analysis of behavioral changes in dairy cows associated with claw horn lesions.
        J. Dairy Sci. 2016; 99 (26874422): 2904-2914
        • Paudyal S.
        • Maunsell F.P.
        • Richeson J.T.
        • Risco C.A.
        • Donovan D.A.
        • Pinedo P.J.
        Rumination time and monitoring of health disorders during early lactation.
        Animal. 2017; 12 (29143705): 1484-1492
        • Russell R.A.
        • Bewley J.M.
        Characterization of Kentucky dairy producer decision-making behavior.
        J. Dairy Sci. 2013; 96 (23628248): 4751-4758
        • Rutten C.J.
        • Velthuis A.G.J.
        • Steeneveld W.
        • Hogeveen H.
        Invited review: sensors to support health management on dairy farms.
        J. Dairy Sci. 2013; 96 (23462176): 1928-1952
        • Schirmann K.
        • Weary D.M.
        • Heuwieser W.
        • Chapinal N.
        • Cerri R.L.A.
        • von Keyserlingk M.A.G.
        Short communication: Rumination and feeding behaviors differ between healthy and sick dairy cows during the transition period.
        J. Dairy Sci. 2016; 99 (27720146): 9917-9924
        • Siivonen J.
        • Taponen S.
        • Hovinen M.
        • Pastell M.
        • Lensink B.J.
        • Pyörälä S.
        • Hänninen L.
        • Fogsgaard K.K.
        • Røntved C.M.
        • Sørensen P.
        • Herskin M.S.
        • Haley D.B.
        • Rushen J.
        • De Passillé A.M.
        Impact of acute clinical mastitis on cow behaviour.
        Can. J. Anim. Sci. 2011; 80: 257-263
        • Singh Y.
        • Lathwal S.S.
        • Chakravarty A.K.
        • Gupta A.K.
        • Mohanty T.K.
        • Raja T.V.
        • Dangi R.L.
        • Roy B.K.
        Effect of lameness (hoof disorders) on productivity of Karan Fries crossbred cows.
        Anim. Sci. J. 2011; 82 (21269376): 169-174
        • Sørensen L.P.
        • Bjerring M.
        • Løvendahl P.
        Monitoring individual cow udder health in automated milking systems using online somatic cell counts.
        J. Dairy Sci. 2016; 99 (26547650): 608-620
        • Soriani N.
        • Trevisi E.
        • Calamari L.
        Relationships between rumination time, metabolic conditions, and health status in dairy cows during the transition period.
        J. Anim. Sci. 2012; 90 (23255819): 4544-4554
        • Stangaferro M.
        • Wijma R.
        • Quinteros C.
        • Medrano M.
        • Masello M.
        • Giordano J.
        Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part II. Mastitis.
        J. Dairy Sci. 2016; 99 (27372584): 7411-7421
        • Stangaferro M.L.
        • Wijma R.
        • Quinteros C.
        • Medrano M.
        • Masello M.
        • Giordano J.
        Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part III. Metritis.
        J. Dairy Sci. 2016; 99 (27372583): 7422-7433
        • Stangaferro M.L.
        • Wijma R.
        • Quinteros C.
        • Medrano M.
        • Masello M.
        • Giordano J.
        Use of a rumination and activity monitoring for the identification of dairy cows with health disorders: Part I. Metabolic and digestive disorders.
        J. Dairy Sci. 2016; 99 (27372591): 7395-7410
        • Steensels M.
        • Antler A.
        • Bahr C.
        • Berckmans D.
        • Maltz E.
        • Halachmi I.
        A decision-tree model to detect post-calving diseases based on rumination, activity, milk yield, BW and voluntary visits to the milking robot.
        Animal. 2016; 10 (27221983): 1493-1500
        • Steensels M.
        • Maltz E.
        • Bahr C.
        • Berckmans D.
        • Antler A.
        • Halachmi I.
        Towards practical application of sensors for monitoring animal health: the effect of post-calving health problems on rumination duration, activity and milk yield.
        J. Dairy Res. 2017; 84 (28524016): 132-138
        • Steensels M.
        • Maltz E.
        • Bahr C.
        • Berckmans D.
        • Antler A.
        • Halachmi I.
        Towards practical application of sensors for monitoring animal health; design and validation of a model to detect ketosis.
        J. Dairy Res. 2017; 84 (28524012): 139-145
        • Tse C.
        • Barkema H.W.
        • Devries T.J.
        • Rushen J.
        • Pajor E.A.
        Effect of transitioning to automatic milking systems on producers' perceptions of farm management and cow health in the Canadian dairy industry.
        J. Dairy Sci. 2017; 100 (28109587): 2404-2414
        • Van Hertem T.
        • Maltz E.
        • Antler A.
        • Romanini C.E.B.
        • Viazzi S.
        • Bahr C.
        • Schlageter-Tello A.
        • Lokhorst C.
        • Berckmans D.
        • Halachmi I.
        Lameness detection based on multivariate continuous sensing of milk yield, rumination, and neck activity.
        J. Dairy Sci. 2013; 96 (23684042): 4286-4298
        • Veissier I.
        • Mialon M.M.
        • Sloth K.H.
        Short communication: Early modification of the circadian organization of cow activity in relation to disease or estrus.
        J. Dairy Sci. 2017; 100 (28318584): 3969-3974
        • Walker S.L.
        • Smith R.F.
        • Routly J.E.
        • Jones D.N.
        • Morris M.J.
        • Dobson H.
        Lameness, activity time-budgets, and estrus expression in dairy cattle.
        J. Dairy Sci. 2008; 91 (19038930): 4552-4559
        • Weigele H.C.
        • Gygax L.
        • Steiner A.
        • Wechsler B.
        • Burla J.B.
        Moderate lameness leads to marked behavioral changes in dairy cows.
        J. Dairy Sci. 2018; 101 (29290435): 2370-2382
        • Wittrock J.M.
        • Proudfoot K.L.
        • Weary D.M.
        • von Keyserlingk M.A.G.
        Short communication: Metritis affects milk production and cull rate of Holstein multiparous and primiparous dairy cows differently.
        J. Dairy Sci. 2011; 94 (21524531): 2408-2412