Precision Dairy Technologies – Opportunities and Challenges

welcome everyone to New South Wales dpi dairies webinar number three precision dairy technologies opportunities and challenges this webinar will be a bit different to the previous ones as we have the great privilege of having up top international guest speaker with us here today dr. Jeffrey bewdley from the University of Kentucky in the US has kindly accepted our invitation to share with you a bit of his deep knowledge of precision dairy technologies as an introduction to our guest speaker dr Beauty grew up in Kentucky an alien life form involved in dairy working on his grandfather’s farm dr. Beauty has a Bachelor of animal science from the University of Kentucky a master of day science from the University of wisconsin-madison and a PhD from Purdue University currently dr. bewdley is a researcher at the University of Kentucky and his main area of work focuses on precision dairy technology implementation phosphate is prevention Cal compound name is prevention and decision economics he has numerous publications and scientific journals and industry articles and a spoken at a conferences all around the world dr. Buena – the stage is yours and the privilege is ours to have you with us here today I appreciate the invitation to present you all this morning it’s evening here in Kentucky talking today about an area that I think is very exciting all of the different technologies that are coming on board for the dairy industry and think about what we might be able to do with these technologies to talk a little bit about some of the research that we’ve been doing here and to discuss some of the practical implications of making these technologies work on farm I always want to start out my presentations by thanking my graduate students undergraduate students and farm staff for all that they do to make the work that we do actually happen you think about technologies and sensors it’s nice because these technologies provide us ways for the cows to talk to us in the past we may have bought which the cows could tell me when something was going on and in many cases these technologies are providing that opportunity for the cows to tell us what’s going on with them on a day-to-day basis the area that we’re focused on here in our work is what we would call precision dairy monitoring and precision dairy monitoring is looking at the milk components the conformation or the behavior or physiology of the animal trying to better understand what’s happening with the individual cow and one of the nice things about precision dairy monitoring if it enables us to move from just thinking about managing groups of animals or herds of animals being able to go back to the idea of managing the individual towel one of the questions then is can we get the happy cows through the use of technology and you can see a few different examples here what’s going on at the University of Kentucky we have precision paddy our mascot Theron sees a blue-and-white blue-and-white are our University colors and she’s wearing all the different technologies that we have here at the University of Kentucky you can see in image analysis representing some of the image analysis work that we’re doing the wearable technologies leg based neck based ear based technologies that can provide us a lot of information about the behavior of the animal the happy lab is one example of a technology that used in parlor to monitor mel components and then the end result is what you see there with five pin a cow in our herd who actually the smiley face on her teeth when I look at the cow I think about all the different possibilities of what we could measure on that count and there are a number of different items that we may be able to measure on count it’s actually amazing to think about all the information that this cow can provide US and many of the items that we have listed on this screen are things that in some fashion or another you’re a dairy farmer you’re measuring or monitoring in some way or another on your farm today maybe not all of them do we have the ability to measure automatically or record that information but we’re at least looking for many of these parameters now when I look at that list also start to think about a practical reality and that we probably don’t want to actually measure all those things we can rain to a trap of falling in love with the information and thinking that if we could measure this wouldn’t it be neat and we can measure more than what we actually need and in many cases I think that goes into a situation where it’s not economic I think it’s quite clear that if we measured all the parameters on the screen for our towels that would not be economical to do that

and certainly I would say that even here at the University of Kentucky because we’re research-based institution the value of the technologies on any single one of our cows is more than the value of the actual cow and that’s not an economical scenario in economics is something that really it absolutely impede through what we’re doing with these kinds of technologies there are a lot of different potential applications of precision dairy monitoring estrous detection is probably the most common way commonly used technology we see a lot of farmers that have invested in estrus detection systems around the world these systems do work they do a really job really nice job picking up cows in heat we see systems that could be used for mastitis detection for fresh cow disease infection lameness defection having detection potential for measuring and monitoring things that we could use this new genetic trait being able to measure phenotype that we previously might not have been able to measure and being bringing that information into our genetic evaluations so that we can actually select for traits like locomotion or body condition score and then using this information at the pin or the or the herd level for managing what’s going on within the herd and monitoring changes across time or compared to other operations there are a lot of different potential precision dairy benefits one of these I think the one that we’re the most interested in here is improved animal health and well-being early detection of conditions increased efficiencies improved product quality minimizing the environmental impact of the theory and providing us more objective measures of items that we’ve typically measured subjectively things like locomotions go a body condition scoring that we’ve always done subjectively perhaps with technologies we can measure these things effectively and get more reliable regularly recorded information that we can use to manage our dairy operations when we think about disease detection it’s quite logical to think about how it may be beneficial to detect disease early if we detect disease early we might be able to treat the animal sooner we can improve our prevention program and hopefully these these actions will help us improve the outcome from the treatment which is production losses produce economic losses and improve animal well-being and our operations now it’s important I think stop here and recognize that when we say these things it does make a lot of sense that this could be the case but if we actually try to quantify how much is the value of early detection of early treatment very difficult to do that and there’s not much in the literature demonstrating what the value of early disease detection actually is if you look across the market it’s absolutely overwhelming how many different options there are there are so many technologies that we can now use on our dairy farm and if you’re sitting on the farm today it’s a difficult question to think about is that if you decide that it’s time to invest in a technology which technology should you invest them I think it’s it’s really difficult to decide with so many options so many different strengths and weaknesses of different technologies and where they are in the development stages what are the criteria that we should look for in a technology here’s the criteria that I think we would look for in an ideal precision dairy monitoring technology first of all I think that technology has to explain an underlying biological process it has to explain something that’s going on within that animal it has to be something that we can turn into a meaningful action needs to be a piece of information that not just knowing what it is but by knowing what it is I can make some change or actually relative a particular animal or group of animals to make it different with that situation the technologies have to be cost effective that doesn’t necessarily mean low cost but we better be able to demonstrate an economic for using that technology the technologies have to be flexible and they have to be robust and reliable it’s a very difficult environment that we ask these technologies to survive in we have big clumsy animals working in wet dead dirty environments the technologies have to be simple and they have to be solution focused many times I see an engineering gadget that’s looking for an application but me I would much prefer scenarios where we have an issue on the dairy farm or an opportunity on the farm and we find the technology that can we can use to help manage that well starting with the cow and moving towards the technology instead of starting with the technology and moving toward the cow lastly I think that the technology has to provide information in a format that’s relatively easy to understand that’s a simple way of combing through all that data making some kind of an alert list some kind of a dashboard that provides us the key pieces of

information that we need to use within a day our dairy operations let’s take a look at some of the different technologies that that are out there we’ll start in the parlor one of the most exciting opportunities in precision dairy monitoring is the systems that are now coming out do in-line somatic cell count we have multiple options now that are on the market or coming on to the market that can provide as individual somatic cell count for individual cows each milking for a mastitis management perspective that’s early detection tool I think it’s pretty exciting on the other hand I think we have to be cautious because I think there’s a camp that we actually may use a technology like this to increase the number of treatments that we have and if we increase the treatment and that’s treatments that are not necessarily needed you would possibly increase our antibiotic use which may have an impact on antibiotic resistance but more importantly we increase our dumped milk and that has a direct impact on the bottom line because that costs the same amount to make but we don’t get as much from it when we have to dump it and one of the challenges is that when we’re measuring this frequently will actually find that there are cows that periodically spike in somatic cell count or a mounting or two and then they drop and they go back down to normal and the tendency on that might be to treat that couch he has a high somatic cell count let’s be proactive let’s Peter but in many cases this is actually just a natural reaction of a healthy cow the cow has mounted in the mean response and she beat the infection and this happens every day in our dairy operations but we miss it because we’re not measuring that so I think we really have to be cautious about actually over treating with these inline somatic cell count tools there multiple systems now on the market that used spectroscopy which is basically looking at the pattern of light flowing through milk the multiple arrays of light that those would look at it’s an indirect way to look at changes in milk composition they happy lab for example uses near-infrared and it measures fat protein and lactose in line the nice thing about this kind of technology is that there are no reagents involved there’s no variable cost associated with using this technology across time another exciting option coming onto the market it’s a bit of a herd navigate this system measures multiple parameters in the milk and measures progesterone for example which is a good measure of heat detection can be used for pregnancy infection can be used for detecting cows that are cycling or not cycling can also be used for abortion detective really a powerful powerful tool for reproductive management it goes from looking at indirect measures of estrus for example like activity a direct measure of what’s going on physiologically but then that animal this technology also measures the LDH enzyme which is an early indication of mastitis it measures bhp a which is an indicated of something of ketosis and it renders urea the multiple parameters measured within the same system to a vine as a tremendous amount of information about what’s going on with that particular animal we look at the wearable technologies I think it’s interesting to compare what’s going on in the dairy industry with what’s going on in the human industry we see a wide adoption now of parameters of technologies that we can wear ourselves the measure of what’s going on with our with our physical activities and our health now when I talk to people who are not in the dairy industry about what I do it’s very easy for me to describe that we work with fitbit’s for cows and these are exactly the same idea we’re using the same base technology pedometer technology measure activity in the animal there are multiple systems in the market now that measure ear or neck based behavior these systems as the name indicates are sitting on the neck of the year of the cow they measure activity they measure things like rumination time and eating time or grazing time these technologies provide us a tremendous amount of information about what’s going on with the cow on an individual date and I think can be very powerful tools for managing dairy operations there multiple technologies that measure physiology we may take for example a monitor that measures vaginal temperature we have a technology the fever alert there that measures ear canal temperature or we have multiple room and vollis’s on the market that measure temperature each of those unfortunately comes with its own set of disadvantages vaginal temperature is a limit because we can’t leave a device in the animal for a very long time without creating an infection rhythm and temperature is limited because every time a cow brings water there’s a dramatic decrease in temperature that occurs and it can take as much as three hours before that temperature returns back to normal therefore there’s a lot of information

that’s provided it’s not really representative of core body temperature a system like the fever alert there is a great concept we can get a temperature that goes into the ear canal we get through a panic temperature which is a really good temperature raise is very close to the hypothalamus however the challenge with its kind of a device is kicking it in the cow we use this system for a while here at the University of Kentucky and eventually had to think about because we couldn’t keep the cows keeping it in their ears there’s also a system that’s that’s out now mainly used for horses but it’s called check the vet check system it measures respiration rate blood pressure and heart rate and then we have a system like smack stick bolus their representative of multiple voulez’s’ that are out there now that measuring mph room pH can be a very powerful tool for monitoring ruminal acidosis one of the challenges though what this kind of technology is what we call sensor drift the technology may be measuring pH today but if we go back and look at it two or three weeks from now the pH may not be the same measurement that it was before because we don’t have the ability to go into that cow recoup that device and calibrate it each time each day there are also multiple technologies on the market for measuring line behavior buying behavior I think is a very useful tool for unfarmed evaluation of line time it can provide as an identification of cows that requires some attention whether that be lameness illness or estrus it can be a good way to assess facility function or cow comfort and it can be a potential metric used to monitor and lobby then we have real-time location systems these systems are basically GPS systems for cows with these systems we’re using something called triangulation which is identifying the strength of signal from the Fang on the cow to three points around it and using the strength signal signal to each point identify where that animal is I think these technologies could be very interesting from a day-to-day manager perspective because I think these technologies will provide us some valuable information in seeing how the animal changes pattern of where they spend their day this may be an indication that they are sick or in heat but I also think that these technologies are really potentially very useful are very exciting for being able to look at the cows response to her physical environment being able to use this information to better understand where cows spend the time when they spend the time there and using that better design facilities and better to determine when we could make some micro environment changes so if for example we may find that animal in one particular group or more heat stress to the other and we may need to give them supplemental cooling more than what we need to give the other animals they’re also quite a few devices on the market now for calving detection one of the ways that we can do this is with a probe that goes into the vagina and these bows measure temperature there is actually a 20 but there’s actually a huge drop in temperature about 24 to 48 hours before calving let drop and temperature is used to identify approximate calving time there are also a couple of new monitors on the market that are using a tale based approach it’s basically like a wristwatch that goes around the tail and this system is monitoring for hell rising weight does the cow gets closer to parturition he starts raising her Delmore and this technology is going to pick up on behavior we can also work here at the University of Kentucky on developing some new technologies one area that we worked quite a bit in is the use of 3d body condition scoring automation basically using an image analysis a camera here in this case it’s the Kinect camera which is actually the same camera that’s used in the x-box and we’re taking a 3d image of the cow we’re taking that from above and looking at the contours around the hooks around the pins and around the head to identify deposition of fat to give us an indication of body condition score the Laval also now has a system being commercially marketed to do this we’ve also played around a little bit with the idea of measuring feed intake using image analysis so here you can see multiple feed bends one of them in the top middle is is completely full the others have some feed missing from them and what we’ve done with that was we were able to identify the change in volume based on the contour of the feed and that would be an indication of feed in thick but also looked a little bit of sleep there’s not a lot of work that’s been done on sleep in dairy cattle but it’s something we’ve become interested

in as we’ve seen more of our herds moved to loose housing systems the stall basic systems we’ve seen those animals sleeping more than what we thought we had seen before we know the importance of sleep in humans we know that it helps with immune function we know helps with our well-being they know it helps us resist disease but potentially we could use this information for the same things along with building a better barn for the cow if we understand rest quality through monitoring of sleep instead of just rest quantity and the study that we did was a proof-of-concept study and we found that 90 to 93 percent of our observations agreed with human observations for and the cow was actually sleeping we have worked with multiple technologies for validating them there was a particular study that Matthew Borchers did where we used five different technologies on the same set of cows and monitored multiple parameters from those technologies you can see here from the line time analysis that our correlations were one meaning there was a perfect correlation between the eggs indication of line behavior and when our observers indicated that the cows were lying down we’ll also looked at a couple of different feeding behavior monitors and in both those cases those were very good results indicating that we predicted a high percentage of the variation in feeding behavior and can relatively our visual observations the planation time was not quite as good for one of the technologies however was quite good for the other technology dimension we’ve licked some that ruined pH in fact I think we worked with four different ruined pH bonuses now here’s an example of what we see one of the challenges is you can see in the blue bars these are the probe measures where we put the probe and measure the rumen fluid pH and the red indicates the the bolus pH and there are some significant differences there between them I think part of what’s going on here is that the system is just over us we under estimating pH and that’s something that can be relatively easy easily directed for but gives us a little bit of doubt maybe about what’s going on with the room pH the other thing is we don’t know exactly where that room and pH reside where that room and bullets resides within the animal it could have something to do with where the bolus is sitting compared to where the pH sample was taken out of the animal but this is probably the biggest biggest concern we see with rumen pH monitors on the y-axis here on the graph you see this is the difference between the probe and the bolus and in the beginning you were very close but across time we start seeing more of a difference between the probe and the bull is indicating that since it does drift across on the other hand we were able to use this technology in a an acidosis challenge in this particular case we we challenged these animals with acidosis basically giving them a lot of grain and you can see that I monitored pick up a significant difference and rumen pH on the day of challenge knowing that we can potentially pick up acidosis with this kind will also work along with estrous detection technologies this was an interesting study that Lauren Mayo one of my graduate students conducted here where we looked at multiple devices on the same animals we synchronized these cows and on the last job we get velocity and RH shot allowing them to come in during in natural heat and we monitor the cows four times a day 30 minutes each where we were doing nothing but visual observation of esters the interesting thing here from this slide is that across the technologies there aren’t any real differences and they built of these technologies to detect the cows that were in heat the third one from the left was a little bit worse however that’s not one that’s available in the United States today so all these technologies did a really nice job of identifying one the cows in heat and the other thing that was neat was that they all indicated that she was in heat over twenty to thirty percent more times more than what we did so these technologies are better than humans at picking up cows that are in heat another study we did on my understanding you guys probably don’t do quite as much synchronization is what we do here but we compared two different sets of herds three herd where we had half the cows go on to a synchronization protocol a ki 7g protocol which if you don’t know is a very aggressive protocol and the other half of the cows we use an activity

management system to know when to breed those cows and the results were quite interesting we found was that there weren’t any differences for most of our reproductive parameters at the end of the study which would have been for the middle of ligation so in this case we can see that there are no differences between proportion of the cows pregnant at the study there were no differences in days open pregnancy lost services for conception or service conception rate however there were two significant differences for the time the AI system the synchronization program you can see that there was a lower days to first service for the found AI system that makes sense we are intentionally waiting and reading everybody on a specific day in milk the opposite of that is the service animal and you can see in this case it managed bills the other way the activity management system has a an advantage for the service interval these cows we don’t have to wait for another pregnancy check and no one to read them again though between the two factors there that are different that essentially Wasi watch each other out and we end up in the same place whether we use the synchronization program or the activity management system we’ve also worked a lot with disease status so we worked with monitoring multiple diseases simultaneously it’s just some work from an Amanda stone one of my graduate students here where we looked at the changes that occurred around mastitis with mastitis we saw about 110 decrease in minutes per day ruminating and so on any one difference in neck activity a five kilo or ten leader ten pound difference in Mel killed and a half degree difference in particular room and temperature all these were significant predictors of mastitis I will warn you however that when we ask the question the other way if we have the alerts how useful are they we may not see quite as good a results there and this is a graph demonstrating the pattern of fat protein ratio in two groups of cows the group in the red where Charles de had subclinical ketosis and the glue that group in the blue for cows that did not have some clinical ketosis and you can see from this starting about day three the fat protein ratio starts increasing dramatically for cows worth of clinical ketosis compared to cows that do not have subclinical keep those this one really blew me away you see here we have days and milk on the y-axis and in milk on the x-axis the orange line represents cows without subclinical hypercalcemia and the blue line represents cows with subclinical hypocalcemia in this case there were over 20 pounds or 10 kilograms difference in milk yield across that period of time it’s a huge economic loss for our farms not to mention the other effects of these diseases here’s another example of an exciting possibility this is the practical system it’s made by EMGs another company based in israel in this case we can see that there was almost an hour difference in the amount of time that the cows spent at the feed bunk for cows that were subclinical versus not subclinical hypocalcemia because you have to talk about the economics economics is actually what first got me into this area of precision dari I was very interested in the economics of decisions and was looking for a dissertation topic and this was what I decided I wanted to work on I really do think that from an economic perspective with the size of the investment that we’re talking about making here we need to do some investment analysis we need to do more than this just feels like the right thing to do or all my neighbours have done it so I need to do these systems are not necessarily one-size-fits-all but makes sense for one person may not necessarily be what makes sense for another one it’s really most easy to show the economic benefits for heat detection and for reproduction I think we can find and agree that systems that measure multiple parameters are the ones that make the most effects this is an example of the kinds of economic work that we’ve been doing this is a dashboard tool that carmella dolly’ egg one of my graduate students developed farm can go in and put their individual farm specific information and what it

will end up with is a net present value analysis of potential technologies you can see this is based on estrous detection the technology on the right goes to be a positive investment with a break-even time of a little bit over five years the technologies on the Left were not breakeven indicating we should not invest in those two particular technologies and that particular dashboard is available on the internet so if you’re interested in looking at estrous detection or your herd and looking at whether or not you should invest in estrus detection technology i encourage you to check out that particular website of course no matter how much we model there’s always an intangible value to having information this is a picture from a herd that I worked with quite a bit over the years in this particular day that we were here the cows were very hot in this barn and we had some County Agricultural agents with us who indicated that they thought that was the case and they started asking the farmer why and he said you know now that you say that I’ve noticed that the cows with the rumen bolus they’re immature in this barn is always a lot higher than for cows that are in the other barn why do you think that is well he did not have a ridge vent on his barn to allow air to escape and he didn’t have enough fans so we mentioned to him that maybe he should add a ribbon and we measured how much when he was getting with those fans and we determined he means more fans what’s the value of that this farmer actually made those changes we’re out there on a Wednesday by Monday morning he had texted me this picture join me the changes that he made it’s very difficult to model that kind of change so that’s one of the challenges that we have in looking at the economics I think it’s important that we keep a balance between the technological transformation and all the associated challenges associated with using a technology you can see here this is a fiction of what happens with a product across time we start out with a technology trigger this is when a new technology comes to market and we start learning about it we see the ads in our magazines then we enter in this period of inflated expectations we start thinking the technology is going to be able to do more than a technology can do after that we enter into a trough of disillusionment this is basically where we start getting frustrated because we found out the technology doesn’t do everything that it was supposed to do it doesn’t do everything that we wanted it to do and after we get all that we go into an Enlightenment period and a productive period where we actually make these technologies work it’s better than the period where we were disillusioned with the whole deal it’s also lower than the peak of inflated expectations to me this graph looks together very well what oftentimes happens to us with these kinds of now in my mine with Esther subjection technologies we’re on the right we are actually where we’re seeing these Esther’s attack work well we know what they can and can’t do and they can be a great tool to add to the operation on the other hand I think with many of the health detection technologies that we’re still in the peak of inflated expectations still not quite where we need to be with regards to our expectations for how these technologies might work and what their limitations might be sometimes these technologies are marketed as plug-and-play but maybe they’re really more plug and pray or plug and pay this is an example of our dairy herds office really not too far from the truth we have multiple computers multiple monitors all around that room why because the technologies indicate the technology companies indicate that they do not want their software on another computer with software from another company that’s very difficult to communicate with the herd management information very difficult to get information out of these technologies so this is one of our big limitations as an industry is communication of data across multiple platforms as a dairy producer I’m not sure how we want to go through eight or ten different computers to understand what’s going on in my herd today I’d like to be able to go in and have some kind of an indication that particular day what’s happening on the farm so I can better understand how to manage that situation another thing we had to be cautious of is that we didn’t do see this kind of approach to marketing the disease detection technologies so a picture and we say look at this cow she was lying down six seven hours a day for a while and then she started increasing things

along the 29th of November he was actually lying down over ten hours a day so what we do is we say we identified this cow head nice tire is there for the technology could have helped us pick up when this cow had mastitis and that’s good and it shows some of the potential these kinds of technologies but the other side is there are carols that show that kind of spike they had nothing going on and there are cows have a almost perfectly straight line for the behavior and they had a lot of things go along so the sensitivity and the specificity of these technologies is not necessarily very high so let’s bring this to the points of sensitivity and specificity but sensitivity basically what we’re looking for is how many cows with the condition do we actually find well if I have 100 estrus events how many of those estrus events is the technology by in the example here I’ve identified 80 ester signals but the flip side of that is I’ve missed pointing it’s that good enough well everybody probably would say they wanted to be a hundred percent but that’s not realistic so I would say eighty percent sensitivity twenty percent specificity actually pretty good now the other side of that is if I have a hundred alerts how many of those alerts are for cows that actually have something going on with them so if I have 100 alerts and ninety of them are for cows that are actually he the ten of them are for cows that are not can heat is that relevant it’s that what we want to see I would argue maybe we want to see a little bit higher because the number of alerts that we can see on our alert list every day if we manage it in this way can be quite high this will be a 90 percent specificity so the scenario here that we have is an eighty percent sensitivity for the 90 percent specificity that’s pretty good now where we should be depends somewhat on the individual farmer it also depends on how much they think that they may have in terms of reproductive problems associated with using these technologies more often we have a lot of questions that we need to think about about how we handle the data what’s our protocol for handling alerts do we examine Carol’s be Carol’s how we learn when we don’t treat the cows what did we do about natural reactions of healthy cow that’s somatic cell count example that I used earlier what are we going to do in that situation I would argue probably the best thing we could do is alter her but if she does break with penico mastitis we’re better better able to go in and handle it what are we going to do about repeat alerts cows that show up on the alert list every single day how do we get rid of those how do we hide those from the farmer hi what about failed devices not something that’s talking about very often but devices do fail we would say that it would be a good reasonable goal to assume that we can get to three percent failure rate within a year which is great on the other hand we’ve seen technologies that said that they will work for seven or eight years and then they don’t work as well so we have to think about those failed devices we’ve seen situations where we may see 10 percent failure rate for a week I think what we’d like to see is as a 2/3 percent failure rate per year and what are we going to do if there’s a system outage if I’m relying on the system to to monitor my cows for health to monitor for asterisks then what am i doing – if the system goes off lightning strikes tristan goes out we need to have a plan for how we’re going to manage our cows in those situations then we look at how we can use the data at the group or the cow level i think that data is most useful for within-group changes or in her changes it may also be useful for cohort comparisons when we use this information we need to keep in mind natural variation and lag and that’s information lags quite a bit behind it’s not representing what’s happening today it’s representing what’s happened over the last month’s like body condition score for example and natural variation is is there where we’re going to see differences from day to day in things like rumination time or feeding time or grazing and we have to be careful not to react with a small change in a small period of time we need to be really careful and cautious preparing information across herds it’s really easy to get caught up in the my rumination time is lower than yours or higher than yours and say one’s better or worse than the other but it’s not really that straightforward we have to consider

production levels at her we have to consider how the tags are attached on one particular compared to another you need to be really careful in doing benchmarking across her and we need to question conventional wisdom any question things that we’ve heard for years we hear that if cows lie down more cows room anymore they’ll milk more and you can see here that there is a positive relationship between now Kilda rumination time but it’s not as strong as what we necessarily would have thought it would be we can see that there is a decrease in line time associated with an increase in milk you that doesn’t make sense for years we’d say that if we get our cows to lie down more they’ll melt more but the thing is counts that melt more have to eat more but not necessarily the same relationship that we would have thought wouldn’t necessarily come up you can see a picture here of my friend David Corbin Dave’s a dairy farmer here in Kentucky we worked with David on a lot of different projects years ago we were getting ready to the technology project I had picked David to be on that particular project and the technology company indicated to me that I made a big mistake they said this guy will never be able to figure out how to use a technology you picked the wrong guy he can hardly use a mouse well as it turned out today he’s super user he is a pure cow guy he understands cows cow behavior cow husbandry and he’s used data from that system and make him an even better cow manager than what he was before he’s exactly the kind of person that’s going to benefit the most from these technologies people that are looking for it as a replacement for poor management are going to be severely disappointed you have to understand the cow to get the most value out of these technologies certainly mistakes happen and we’ve made plenty of them here in our work and simple things that we can into overlook we don’t think about them first of all he did excellent systems only get cows that are in heat they do not catch cows that are not cycling if a system takes up a sick cow she’s still a sick I’ll wait in the focus sometimes so much on detection that we forget that we’re picking up something that’s wrong and so maybe we we over compensate for that we focused more on detection than we should and not enough on prevention we don’t do anything with the information it was absolutely useless and sometimes really guinea pig and then we have to go through some growing pains and working with these technology we have learned here at the University of Kentucky that cat5 cable is a raccoon delicacy raccoons love to eat cable they eat cable they move cable they employed things they are our biggest enemy for keeping our technologies working here at the University of Kentucky and they say that lightning will not strike the same place twice but I can assure you it will strike but same technology twice lightning is a big problem for us and it can knock our systems out for a weeks at a time whole area of precision dairy still new we’re still learning we’re really at the beginning of this movement when we use this information we didn’t make sure we have a goal for how we’re going to use it let’s don’t just collect a bunch of information and figure out what we’re going to do with it later you need to be careful not to jump to conclusions we can look at trends and and broad conclusion that’s not really there and we need to be careful I’m not trying to over interpret that information and we need to remember that no matter what some information it’s just useless in fact probably more useless information and there is useful information within a technology these are six things that I think people should ask before they invest in a technology first of all what’s the sensitivity and specificity of the condition of interest if I’m trying to detect asterisks or mastitis how well does the technology actually do for picking those things up what percentage of the devices fail in a given year what’s the warranty policy what’s the policy for upgrading to new versions of devices making sure we capture all the full costs the hardware cost device cost maintenance cost and data storage costs and then making sure that we can get in touch with existing users it’s really important I think for farmers to talk to other farmers I’m trying to understand how these technologies might fit within their operation I get questions all the time from people asking me what’s the best technology what’s the one that I should invest in and I will always say the same thing where you can get the best customer service you want somebody that’s going to be able to help you through the installation process helping you learn how to use the technology and something that’s going to be there for you as you work the problems and challenges with the big knowledge where are we going from here I think we’re

going to continue to see more sensor based systems over time I think we’ll see more of these that are milk and image-based and maybe a little bit of a trend away from the wearable technologies will see more technologies focused on animal well-being and environmental impact maybe even more so than reproductive health see more multi parameter systems more use of machine learning things like neural networks and fuzzy logic helping us better understand how to use the information coming from these technologies looking at individual farm algorithms not assuming that every farm responds exactly the same need to adjust these for a specific bar we’ll see more open source hardware things like the Raspberry Pi more cloud based data integration or user groups farmers getting together to talk about using the technologies we’ll see more demand for quality alerts and decisions made beyond just it feels like the right thing to do look at this area a big data and I think it’s really exciting it’s exciting to think about all the different things that we can measure all the new ways that we can manage our dairy operations but at the same point we need to maintain realistic expectations not going to change the world it’s not going to change the the whole systems but it’s going to change some of the way that the cows and the people on the operations interact I sure do appreciate the opportunity to speak to you all this morning you can see my contact information here you have any questions about the technologies please feel free give me a call give me an email message also if you’re interested in this area please follow precision patty she’s our mascot here she has multiple social media outlets that she uses for disseminating information that’s coming out from the University of Kentucky and all around the world about precision Dairy Management thanks again and if we have time I’ll be happy to take a few questions Thank You Geoffrey for such a fantastic insight into this more than exciting topic I think you have been able to give us a very good overview of all the opportunities and challenges that precision dairy farming can have I think you’ve covered a lot of the key items that we’ve discussed here in Australia that these are tools but the key will still be in the management the manager will still have to be in control of everything that is that a lot of these systems provide data but it’s about turning that data into actual information and I think it’s about being realistic about the expectations because I’ve seen more than frequently than some farmers have unrealistic expectations of what the technology can deliver and of course that ends up not delivering up to those expectations and one of the questions that somebody asked through the chat box was um looking at the technologies that are not on the market but but could be coming in the pipeline and you’ve covered a bit of that trying to look a bit ahead what do you think might be a truly disruptive technology solution that might transform dairy farming so it’s thinking ahead and not being constrained by current technology availability but it’s thinking more broader of what things may be I’m not here today but could come in the future it’s hard to predict future I don’t have a very good crystal ball for that for me if there were something that could measure intake then that would be transformative and I think whether that’s a confinement place system or a I pasture based system it’s the same challenge and I don’t think there’s a great way to do that in either system that I’m aware of right now but that would be transformative do you think we are far away of being able to assimilate in take on a cow level in a practical affordable way on and I can find an operation I know we’re far away and and I don’t I’m not as familiar with the pasture based options but from my understanding there’s still some challenges there yeah associated with that no that’s a good one probably I had one from from previous M worker discussion that I’ve had with you but maybe to share with the audiences and how far are we from from going from the actual raw data that the sensor can capture to actually delivering information of value for the farmer and maybe coming up with some solutions or recommendations but the farmer can take given those those parameters that are measuring on-farm

and the second question would be M technology has been available sorry for for some time already but we’ve seen that it hasn’t changed dramatically the way farmers take decisions and that adoption has not been that big what a bit of the reasons for for slow adoption that you’ve seen all around the world well I think one of the challenges with the information is that you know it has to be something that we can do something about and with asterisk that’s one of the reasons why it’s it’s been so it’s been more widely enough than other things because it’s quite clear what we need to do when the counts in heat and so it’s a very dramatic change and whatever we’re looking at behaviorally or physiologically it occurs over a short period of time and there’s a clear action to take unfortunately when we get into some of the health information it curves across time and it’s not as clear what we need to do and so that makes it a little bit more difficult so I think that there there’s a lot of work to be done in figuring out exactly what the best thing to do what the information is but I think there’s also a lot of work that needs to be done in improving the algorithms associated with providing alerts a lot of the systems now are still based on on sort of a simple percentage change or number of standard deviations different from an average and and I think that’s a little bit too simplistic for some of the conditions that we’re talking about the other thing that I think that we run in who is that sometimes the technology may actually be picking up things that we’re not capable of picking up so we may call it a false positive account it wasn’t really sick when we went and looked at her but maybe it’s not really a false positive maybe it’s a subclinical condition that we can’t see or maybe she’s appear to from showing physical signs of what’s going on with with whatever condition the system’s picked up on and and what do we do in that scenario did we do we ain’t can action with that animal or not it’s not so clear what do it yeah that’s that’s what there’s a very good point and just for for the audience that is listening this webinar and you’re coordinating the big data dairy discovery conference in the u.s. in November and related to that topic of integrating data and using analytics tools M do you think there is any value in cloud-based services that integrate data from either multiple sensors at a farm level or multiple sensors across multiple farms and and do you think this is something that will benefit only the farmer or the other will go even across the the chain let’s say from the farmer like when we talk from the paddock to the plates and like connecting the farmer with her with a consumer that those integrations can help more than the farmer I think I think about more at the farmer level but there are potential values for the entire chain even going back to the consumer there are some questions that not with regards to data over ownership and and in oversight there and I think a lot of farmers it makes them quite nervous the idea that consumers would but have that information and I can’t say that I completely blame that fear oh that’s one of the there’s an opportunity but also I think we have to be very careful and and taking it all the way down the chain excellent okay I don’t want to steal any more of your time Jeffery and on behalf of everyone here today I would like to thank you for your time and I really look forward to hear from you again in the future so thank you very much for your time and your insights into this very exciting topic thank you for having me have a good day for more information I encourage all of you to visit the New South Wales Department of Primary Industries website where you will be able to access tools and resources about the dairy industry there is a specific section on robotic milking systems which includes a lot of information on not only robotic milking but also precision dairy farming and previous webinars that we’ve held this webinar will be available at a later time for access so once again thank you very much for joining today and hope to see you next time