What Automated Estrus Detection Systems Can and Cannot Do
1 like319 views
The document discusses automated estrus detection systems in dairy cows, detailing their advantages such as continuous monitoring and mobile alerts, and disadvantages like investment costs and learning curves. It highlights various technologies and methods for detecting estrus, their effectiveness, and factors affecting detection accuracy. Furthermore, it emphasizes the importance of education and adaptation as farmers integrate these technologies into their practices for improved animal health and reproductive outcomes.
What Automated Estrus Detection Systems Can and Cannot Do
- 1. What Automated Estrus Detection Systems Can and Cannot Do Dr. Jeffrey Bewley
- 3. The Options Are Endless
- 4. Changing the Way We Breed • Efforts have increased dramatically • Positive experiences • Only catches cows in heat GEA Rescounter II AFI Pedometer + SCR HR Tag/AI24 DairyMaster MooMonitor/ SelectDetect Track a CowBouMatic HeatSeeker II DeLaval Activity Monitor Anemon Estrus Monitoring
- 5. Heat Detection Aids Records Heat Expectancy chart Breeding Wheel Computer Action Lists Prostaglandins Ovsynch Pre-Synch Mount Detection Aids Kamar Detectors Estrotect Tail head Marking Videotape Heat Detector Animals Androgynous Female Altered Male Precision Technologies Activity Rumination Temperatures Standing Behavior Progesterone Biosensors Mount Detectors Heart rate Feed intake Milk yield Vaginal conductivity
- 6. Visual Detection of Estrus Disadvantages Subjective Time allocation Herd size Duration of estrus Inconsistent Facilities Labor costs Efficiency declines during busy seasons Advantages Cow time Less initial costs
- 7. Automated Detection of Estrus Disadvantages Investment cost Learning curve System requirements Advantages Continuous monitoring Prediction of ovulation or insemination times Individual animal history Alerts on mobile devices Comparable to timed AI No hormone injections
- 8. Estrus Detection Methods NAHMS 2007 7.3 93 40.3 34.7 14.4 5.7 1.4 0 20 40 60 80 100 Other Visual observation Bulls Tail chalk/paint Pressure devices HeatWatch Pedometers Percent Method
- 10. Activity Changes 0 10 20 30 40 50 60 70 80 90 100 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 108 114 120 Activity Time in Hours Estrus
- 11. IceTag™ Activity Monitor Changes on Day of Estrus (11/1/06) for #624 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 16.00 18.00 # Steps ('000) Hours Active Hours Lying # Lying Bouts Longest Standing Bout (hours) Milk (10 lbs) 10/28/2006 10/29/2006 10/30/2006 10/31/2006 11/1/2006 11/2/2006 Acknowledgement: Dr. Oliver Lewis, Chloe Capewell, and Robert Boyce, IceRobotics, Ltd.
- 13. Secondary Signs 0 10 20 30 40 50 60 70 80 90 100 0 6 12 18 24 30 36 42 48 54 60 66 72 78 84 90 96 102 108 114 120 Rumination Time in Hours ACTIVITY RUMINATION Estrus
- 15. Milk measurements • Progesterone – Heat detection – Pregnancy detection • LDH enzyme – Early mastitis detection • BHBA – Indicator of subclinical ketosis • Urea – Protein status
- 17. Variable Changes • Estrus-6 hours before and after first observed standing event • Non-estrus-14 hours before estrus Variable monitored n Estrus Non-estrus P-value DVM bolus reticulorumen temperature (°C) 18 39.29 ± 0.21 38.86 ± 0.18 < 0.01 CowManager SensOor ear surface temperature (°C) 18 24.17 ± 1.20 22.97 ± 0.83 0.20 HR Tag neck activity (units/2 h) 18 61.62 ± 2.04 28.20 ± 0.78 < 0.01 IceQube number of steps (per h) 17 300.82 ± 10.92 79.07 ± 4.13 < 0.01 CowManager SensOor high ear activity (min/h) 18 17.40 ± 0.66 4.25 ± 0.39 < 0.01 Track a Cow leg activity (units/h) 18 321.14 ± 11.87 95.17 ± 7.16 < 0.01 HR Tag rumination (min/2 h) 18 20.47 ± 2.68 32.96 ± 0.54 < 0.01 CowManager SensOor rumination time (min/h) 18 12.90 ± 1.07 22.96 ± 0.57 < 0.01 CowManager SensOor feeding time (min/h) 18 16.93 ± 0.99 8.93 ± 0.65 < 0.01 IceQube lying bouts (per h) 17 0.35 ± 0.09 0.72 ± 0.07 < 0.01 IceQube lying time (min/h) 17 10.19 ± 1.91 24.82 ± 0.95 < 0.01 Track a Cow lying time (min/h) 14 6.56 ± 2.55 18.18 ± 1.81 < 0.01
- 18. How Many Cows With Condition Do We Find? Example: 100 estrus events 80 Estrus Events Identified by Technology 20 Estrus Events Missed by Technology
- 19. How Many Alerts Coincide with an Actual Event? Example: 100 estrus alerts 90 Alerts for Cows Actually in Heat 10 Alerts for Cows Not in Heat
- 20. Machine Learning Technique Technology Sensitivity Specificity Accuracy Random forest CowManager SensOor 100 99 99 HR Tag 60 99 98 IceQube 80 99 99 Track a Cow 100 97 97 Linear discriminant analysis CowManager SensOor 100 100 100 HR Tag 100 98 98 IceQube 100 98 98 Track a Cow 100 96 97 Neural network CowManager SensOor 100 99 99 HR Tag 100 96 97 IceQube 100 100 100 Track a Cow 100 91 91
- 22. Treatment Variable TAI AAM Time to first service (d past the VWP) 6.0 ± 0.2a 17.0 ± 1.2b Probability of pregnancy to first AI (%) 40.4 ± 3.1 41.5 ± 3.3 Probability of pregnancy to repeat AI (%) 41.1 ± 4.1 42.0 ± 4.4 Service interval (d) 42.0 ± 0.1a 24.5 ± 1.2b Pregnancy loss (%) 12.1 ± 2.4 8.2 ± 2.1 Time to pregnancy (d past the VWP) 50.0 ± 2.3 50.0 ± 2.0 Proportion of cows pregnant at 90 d past the VWP (%) 64.9 ± 3.1 66.7 ± 3.1 Automated Activity Monitoring vs. Timed AI
- 23. • 109 lactating Holstein cows at the University of Kentucky Coldstream Dairy • Modified G7G-Ovsynch used for synchronization at 45-85 DIM • Estrus gold standard was verification of luteal regression and ovulation using temporal progesterone patterns and ultrasonography • Visual observation 4X a day for 30min each for 4 days • All cows equipped with 9 commercially available precision dairy technologies Multiple Technology Efficacy Mayo et al., 2015
- 24. Methods 109 lactating Holstein cows at the University of Kentucky Coldstream Dairy January 2014 to March 2015 Cows were enrolled in the protocol 45 to 85 DIM in groups of 6 to 10 cows Observed for estrous behaviors for 30 min, 4X per day, for 4 days Estrous behavioral scoring system (van Eerdenburg et al.,1996; Roelofs, 2005 )
- 25. Activity Changes -50% 0% 50% 100% 150% 200% 250% -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 PercentChange Days Periestrous Device 1 Daily Steps Device 2 Daily Steps Device 3 Motion Index Device 3 Daily Steps Device 4 Daily Activity Device 5 Daily Activity Device 5 Daily High Activity Device 6 Daily Steps
- 26. Temperature Changes -50% 0% 50% 100% 150% 200% 250% -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 PercentChange(%) Days Periestrous Reticulorumen Temperature Max Vaginal Temperature Ear Skin Temperature
- 27. Feeding Behaviors -50% 0% 50% 100% 150% 200% 250% -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 PercentChange Days Periestrous Device 4 Rumination Device 5 Rumination Device 6 Eating Time Device 6 Daily Time at Feedbunk Device 6 Daily Feedbunk visits
- 28. Lying and Active Times -50% 0% 50% 100% 150% 200% 250% -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 8 9 PercentChange Day Periestrous Device 1 Daily Lying Time Device 3 Daily Lying time Device 5 Time not active Device 6 Daily Lying Time
- 29. Lying Bouts -50% 0% 50% 100% 150% 200% 250% -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 PercentChange Day Periestrous Device 1 Daily Lying Bouts Device 3 Daily Lying Bouts Device 3 Daily Avg. Bout Duration Device 6 Daily Lying Bouts
- 30. Percent Changes -50% 0% 50% 100% 150% 200% Lying Time Lying Bout Duration Rumination Time Milk Yield Reticulorumen Temperature Vaginal Temperature Ear Skin Temperature Lying Bouts Eating Time Feedbunk Visits Daily Steps Percent Change at Estrus ParameterMeasured
- 31. Sensitivity and Specificty Estrus detection Method True Positives False Positives True Negatives False Negatives Total Cows (n) Sensitivity Specificity A 76 2 13 18 109 80.9% 86.7% B 72 0 14 21 107 77.4% 100.0% C 45 2 10 34 91 57.0% 83.3% D 33 1 11 46 91 41.8% 91.7% E 51 0 8 6 65 89.5% 100.0% F 35 1 10 15 61 70.0% 90.9% Standing 51 0 15 43 109 54.3% 100.0% Behavioral score 62 0 15 32 109 66.0% 100.0%
- 32. Ketosis Effect -50% 0% 50% 100% 150% 200% Steps per day Motion Index Lying Time Rumination Time Milk Yield PercentChange Parameter Effects of Subclinical Ketosis on Expression of Estrus Subclinical Ketosis No Subclinical Ketosis P > 0.05
- 33. Heat Stress 0 10 20 30 40 50 60 70 80 90 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% MaxTHIatEstrus PercentofCowsStanding Group Date Effect of Max THI on standing behavior at estrus %stood Max THI
- 37. Tabs organize information Description and instructions for user www2.ca.uky.edu/afsdairy/HeatDetectionTechnologies Karmella Dolecheck et al.
- 38. Hover buttons explain inputs and results Inputs adjustable in multiple ways www2.ca.uky.edu/afsdairy/HeatDetectionTechnologies Karmella Dolecheck et al.
- 39. Compare up to 3 different technologies www2.ca.uky.edu/afsdairy/HeatDetectionTechnologies Karmella Dolecheck et al.
- 40. Technology names appear here Net present value shown visibly as either good (green) or bad (red) Black box and “Best Option” indicate the highest net present value www2.ca.uky.edu/afsdairy/HeatDetectionTechnologies Karmella Dolecheck et al.
- 41. Example Analysis $58,582 $63,582 $64,188 $69,188 $94,300 $99,300 $99,906 $104,906 $0 $40,000 $80,000 $120,000 High-100-70 Low-100-70 High-50-70 Low-50-70 High-100-90 Low-100-90 High-50-90 Low-50-90 Net Present Value TechnologyExample Low: $5,000 initial investment High: $10,000 initial investment 50: $50 unit price 100: $100 unit price 70: 70% estrus detection rate 90: 90% estrus detection rateInvestment-Unit Price-EDR Karmella Dolecheck et al.
- 42. What about Non-Cycling Cows?
- 43. • Likely dangerous • Balanced combination of hormonal intervention plus estrus detection • Identify non-cycling cows and intervene? –When to start intervention? –Full synchronization? • Economic and labor considerations of managing both systems? No Hormones?
- 44. • System costs tend to follow razor model (cheap infrastructure, expense in tags) • Initial systems all read in within parlor, but all have moved to continuous download • Small herds likely benefit more from activity systems • Most systems require tags on at least one week before event of interest • Most farmers choose to keep tags on all cows all the time Other Considerations
- 45. • Be careful with early stage technologies • Need a few months to learn how to use data
- 46. 4. What is the policy for upgrading to new versions of devices? 5. What are full costs (hardware, devices, maintenance, data storage)? 6. What protocols are available for handling alerts? 6 Questions To Ask
- 47. Cautious Optimism • Critics say it is too technical or challenging • We are just beginning • Precision Dairy won’t change cows or people • Will change how they work together • Improve farmer and cow well-being
- 48. Path to Success • Continue this rapid innovation • Maintain realistic expectations • Respond to farmer questions and feedback • Never lose sight of the cow • Educate, communicate, and collaborate
- 49. Future Vision • New era in dairy management • Exciting technologies • New ways of monitoring and improving animal health, well-being, and reproduction • Analytics as competitive advantage • Economics and human factors are key
- 50. Questions? Jeffrey Bewley, PhD, PAS jbewley@bovisync.com jbewley@cowfocused.com