our next speaker is Dr David Rosetto Dr Rosetto leads a research program at Iowa State University focused on the development of practical innovations that Advance swine sustainable systems his current research initiatives include maximizing the nutritional value of fet ingredients in sne diets enhancing Animal livability by implementing targeted nutritional interventions and harnessing computer vision AI Technologies to improve production practice and Animal Welfare leveraging his past experience as the technical officer at the Han company he combines invaluable industry experience with academic purs today he's going to talk to us about so PL Innovations breing emerging Technologies and practical Solutions let's welcome Dr [Applause] Roser good afternoon thank you Joanna for the introduction I appreciate the ation by the committee uh it's an honor for me to be here uh discussing with you uh I was invited to talk about my research program and um I joined iate U almost a year ago so I've been constructing the program um I want you to illustrate what the vision for the program is but as we have started we have started constructing a a team of great students and a great group of people around collaborators and targeting key questions for the industry that's my goal is identify what are the challenges key challenges for the industry we cannot tackle all all but we can select those that we believe will have very high impact in the industry so that's my goal today I hope by the end of today you uh you go out with a a Clear Vision um clear illustration of what my our vision is for our contribution to the industry and I want to start by uh for those that do not me uh talk a a little bit about me but especially about the philosophy of My Life um these are the three key aspects of my life that I like to keep it in front of me all the time first set priorities always challenging trying to balance between G family and work but that's a task every day I in the beginning of the day I remind myself what's what my priorities are what's moving my day um the passion for hard work comes after my my dad and then after attending samuro University which has the the is ISL on labor on nain which means work uh conquers everything I'll talk a little bit a little more about that and then service to others and that's what it moved me to to Iowa state from hanor I wanted to increase my contribution to the industry this is uh I'm honored to be part of the industry I really appreciate you guys so that was what it uh it moved me here a little bit about my background so I I am originally from Ecuador but I moved to Honduras to pursue my bachelor's degree in agriculture it's a a fascinating uh University you live on campus and you work on campus so I spent my four years there and I developed my passion for animals in samorano I came to the US first to work as a a farm employee into a South Farm uh so I worked there for two years with a Smithfield company but then I started to pursue my graduate education I went to NC State to pursue my Master's and PhD in nutrition I was very attractive to the statistical program as well so I have a minor there and then uh through the work and research with uh with NC State and and hanor Company I joined the hanor company uh back in 2014 uh so almost 10 years ago where I started working as a nutritionist but then I later um took the charge of technical officer so we seeing nutrition and the technical and the Technologies of the company I like to acknowledge uh great mentors in my life at the beginning of my career uh these were great mentors Dr Jack OD Dr uing my mentor my main main mentor and Dr uh Dean boy they have been crucial into the early stages of my career they continue to be but now I I appreciate being surrounded with mentors so that's andw of the great things I state has offered to me now moving into my program um same as I talk about the philosophy of my life I like to talk about the philosophy of my research program in which we are intended to develop innovative solutions but with the heart of uh scientific understanding of this and then conducting conducting sound research but then at the end of the day because of my back my background bring some apply innovative solution so I'll like to illustrate some of the examples of the work we're doing today and it will be related to these three aspects of the philosophy now as a as an industri and and today and yesterday we also discussed some of these challenges we had there's many challenges that relate to sustainability relate to the market uh we have pressure to act to the access of resources access to water but also the market is is challenging us right into to a reduce uh use of antibiotics uh they are demanding higher quality but yet too we have a pressure for higher feed prices and and lower revenues into our s so that's a big challenge for the industry Animal Welfare is Big too um we can choose any of these challenges we like to focus on those three key uh challenges that I'm listing here especially the post- winning mortality morbidity um the optimal access to welfare resources through the life of the pig and then the S and the pig losses during faring I wish we can do more but these are the areas where we decided to focus our efforts we believe this is one of the uh this bring to the industry greatest opportunities I'll talk U more about this and how we we're trying to tackle each of these questions now so this is a list of some of the questions we have into our lab and some of the projects we are currently working and I group those questions and and projects into two buckets one into what I call it welfare resources that involve access to nutrients access to water access to space um specifically we are investigating a lot more about the feed ingredient and nutrients uh so questions like how much if if uh soy meal becomes more available and and pric uh very nicely into our leas cost formulation how high can we feed that to the pigs in a safe manner also and yesterday we participating a great discussion about the net energy value of soing meal or productive net energy value that's a question to us but we also want to compare that value uh in a commercial versus University setting thinking about uh the difference in and maybe the H pressure pics might have it into the field and that so meal might have have a role there so that's an order of the projects we're working on I have great interest in in working uh more with fiber um I have seen data but we also have work with fiber especially in gestating cells and tackling cell mortality this seems to be important but one of the key areas here is that I don't think we have a great handle on how we determine fiber into our ingredients uh we're still working with ndf values uh neutral deterent uh fiber values I think we have to move to Total uh dietary fiber values that will bring us to an understanding on solubility and solubility of the fiber uh so that's work uh we are collaborating with Dr Amy Petri great work uh she's doing in her lab very honored to work with her in this area and then lastly this is on this bucket is um what is the we we're doing a field evaluation on the access to welfare resources our question with Dr Anna Johnson has been into as pigs C uh increased the grow rate uh and we are keep um we keep our pigs in the same facilities as 15 20 years ago is that having an impact on the welfare of the pigs uh so Emma Alexa here they have done a tremendous job going into the field and evaluating into the field take measurements on the weight of the pigs counting of the pigs space access for space access to feed access to water uh so I expect great resources great results out of that that study now as we talk about Innovative tools and another of the buckets that I like to bring forward for the industry is the Smart farming tools and I just came from an or session on labor efficiency here U great talks about what is coming to our hands there's a couple questions we have one is uh I like to bridge bridge the development and the implementation uh so I have something about that too but we ourselves want to develop uh tools that help us as researchers but help you as producers and the key questions here will be the access to feed and water right after winning uh so there's development we're doing there too uh monitoring grow performance daily for the pigs which then we can apply into Big Data um tools like G jery was showing but more for the production side more for a key a key opportunity I see is the way we sell pics we Market pcks if we do it in a more precise way my experience will tell me that there's anywhere from2 to5 dollar per Peak opportunity if I can be accurate and I can be precise um together with those questions we have to bring uh talented students into our team uh so this is a group of uh two PhD students a master student that started and uh we're fortunate to have great undergraduate help to and uh research assistant uh bimala she joined our lab early this year uh she's pursuing a PhD in bioinformatics and computational biology so you can see where she can feed into development of innovative tools she is in charge and leading the development of the computer vision Technologies Gustavo Lima is pursuing his doctorate in nutrition and but with emphasis in statistics and economical modeling he's working with feed ingredients especially soy meal uh he can get very deep into this and uh and he brings a really really great experience from the field he comes from Brazil but working t 10 plus years um he has a masters in Brazil so very fortunate to have here him here in our team now Nathan he just join um our lab here uh in the summer um I got to know Nathan in my class last year I teach a 425 swine Management Systems class and I pick his interest on data analytics and so I approached Nathan asked about what his plans were and I knew the passion was for data analytics and for pic so we asked if we can combine the two and create a master's program so uh grateful to have Nathan and now Stephanie uh she's been helping the team as an undergrad uh um research assistant for the whole team uh from this summer very fortunate to count with this very talented uh group of people which will move forward um the projects we have so just to give you a couple of examples of the projects we're doing you can understand that these are um on development we just started um we had the first student arriving on on on January but we had the things going um so one of the first question Gustavo had and this is in collaboration with Dr Nick abler is how much Sobe meal can we feed to grow finish pigs so in in award that uh we heard there's going to be a lot more Sobe meal because of the crashing of soybeans um and maybe more affordable um Can can we go beyond 40 50% and feed that pig safely um this is work that Dr gabbler and Dr Greener did in 200 22 and so here are some of the here's some of that information for the diets first diet first phase four phases so first phase the range of soy meal went from 48 to 26% uh 48 being the highest and in your last phase When Pigs reach 105 kilograms by way the range was uh from 13 to a 29% so meal so High um so here is a summary of the results they they saw and I don't know if you can see it in the back uh it's a small numbers apologize but the end we was about 120 kilograms but there wasn't any significant difference in that weight so any of those levels of so meal were fed to pigs and without um an impact in performance so Gustavo came and he asked how much higher can we feed can we go beyond the 4 % and and these will be values that you might only see in the University setting but that's what we that's what you have to do in the University setting to provide information to you as producers in a word that if you can if you can feed Beyond 48% how much how how will you formulate those diets so four phases again and what Gustavo did here is to formulate diets from 26% and very comparable to the first study but he went to 75% so meal so that's pretty high that's into the first phase of the pig and into the last phase um we are into a second phase third phase um by the end of September we'll have finished the project and pigs will be consuming 42% at the highest level so it's it's it's ongoing basis but um Gustavo is one of of those students that like to keep things going and moving fast so he had a sumary for the first phase so this is just preliminary data that I'm showing but just to illustrate the work he's doing uh in the first phase um from 22 kilogram to a 47 kilogram Pig there was a tendency to be uh to the body weight to be reduced mainly driven by the a h a significant linear impact on the Feit intake of the pig but also an improvement in the caloric efficiency as the levels went to uh 75% in this case so um it wasn't the the I guess the uh decrease or reduction in performance it wasn't as drastic as I I was anticipating with 75% soybean meal um we use relative to the discussion we had yesterday on the net energy value of soy meal we still use uh NRC value which is about 78% of the energy of corn in this uh in this study um just to give you a little more background there some of the data we're seeing we'll continue to analyze the data with we PS uh on Monday and I'm pretty sure maybe by next early next week he'll have a a quick summary of the second phase so this is ongoing and and ultimately what he wants to develop is uh regression and broke line analysis for each of these phases uh to guide you as producers if I do see a response in Feit int or body way where is that I where's my um break uh break point right what's my maximum value that I can do so this is an example on the um daily weight gain on the first phase and where he is comparing this is very small apologize is only intended to illustrate but he is comparing different regression lines and broke line U broke line analysis finding a a break point here so that's that's more to come into this project um I like to then transition to smart tools and smart farming tools uh this is the way that I see some of these tool addressing challenges we just discuss right so I'm just going to list here uh challenges relative to sustainability Animal Welfare environmental impact Animal Health qualify labor all of those challenges uh we're facing as producers but I also listed technologies that we are having available to us today and these Matrix is for me just to illustrate how do I see each of these Technologies playing a role for those challenges right you can take a different look at it you can evaluate differently but I think maybe that's a place to start if I have a challenge do I recognize that technology that will help me to address that challenge but then more importantly is if I do recognize the technology how do I evaluate so I take two components what is the technology evaluation and the second aspect and more importantly is the technology adoption right side which if I'm going to adopt a technology I think I need to ask myself if that's going to improve my Revenue so in the case of uh I mention marketing if I can make two or5 dollars more per pck it's improving my revenue is that going to decrease my cost can I save some labor because I have a technology that I can access to ultimately in my system I want to improve the profitability right but the evaluation is is critical as well and is if I recognize a technology like a computer vision um is that computer vision accurate in the way that I expect so it's is estimating the body weight of the pigs accurately for me is reliable if I install that camera into my system meaning that if I expect that daily weight uh measurement for every camera and if I have it for a cycle of pics can I get that 90 90 95% of the times right that's my reliability and then if I have my checkbox the two can I go and implement the technology through my system when I was at hanor we were um debating all the time about Technologies we had 300 plus sites into IA alone and installing cameras or installing sensors in beans Etc it's a big decision right so if I do it I'm I need to have confidence in three in these three key aspects so that's a question I will challenge you uh maybe offer you a matrix like this to evaluate that's something we want to do at I was state to and offer offer that to you offer to developers so then it becomes easier in the adoption side for the producer so here's an example um I'm just taking data from uh my times at hanor on how we'll take a look to Technologies and this was very early on uh into the development of a smart camera this is a pck vision camera there's a lot more today uh but it's just just a way to illustrate how will you evaluate this technology so the three aspects that I will do is the accuracy reliability and implementation of the system we uh we wait um I think this is for two years of data and this is into our research bar where we compared cameras and compared the pens where we had weights too and by the end of the development or testing we had a 90 more than 97% accuracy which for us it was uh acceptable um and you have to decide yourself if that's going to be the case right you expect 99% and how long does it takes you there now for reliability we had at the beginning more issues as you can imagine youall a camera uh the first first first first cycle you might have 98 99% reliability meaning that you have most of the data capture as you expected but but as the Cycles go on you're going to have more problems with electronics with network uh with the cleaning of the camera itself so you have to have a plan maintaining the cameras to make sure you can achieve most of the times more than 98% right the implementation uh for this technology is where we are far away because uh the price is not there yet unless you just take a few cameras to sample your barn um to and if you're going to install cameras in three plus 100 sites um do you need to have a really good plan to maintain replace cameras clean cameras and who is going to do that right so for a research uh testing to evaluate my system maybe uh on an ongoing basis we accepted the technology to be there but to implement commercially know there yet but just as an example of what maybe as producers we can um we can do to evaluate our Technologies now I talk about the development of some of these Technologies uh so here's remala and she's leading the our efforts to develop a computer vision to understand the PA winning Pick behavior um she's interested in measuring feeding and drinking behavior in po winning picks so here U me get through yeah I have that right so uh she this is a a very uh short summary of The the whole effort that takes in the development this is the first time I took it uh so she is a lot more knowledgeable than I am in this area but I think it's is what it takes for you to have a tool right if you want to have a research tool to understand more for our case more of what is happening post winning we are recording videos of the first treat days post winning we're processing the videos right now we are in the phe of the deep development deep learning model for the object detection so she is getting very close to detect the pig now once you have the detection pck then you can do an object track and analyzing the behavior which will be uh feeding and drinking Behavior that's what we are interested and I will tell you why and then our purpose is to feed that information to a machine learning algorithm to predict the behavior Thea beh behind pics that's ultimate goal is if we can recognize early on fall behind pics I think we can alert employees we can understand more about nutritional interventions that can help address that specific question and and then the program or the de uh model will be ready to do a beta testing into a field condition so I talked to you about the complexity of developing of developing these tools this is just a couple of examples on trying to label and put the right label into the Pig that is getting to the feeder and or getting into the Drinker and and label as a pck standing pck uh eating pck drinking and there's things like augmentation that you can do to feed more information to the m the Deep learning model as I understand it just to get closer and faster to that development right so that's something ongoing and the reason on why is because we're interested into the early Feit intake of that pig uh so this is work we're collaborating with Dr Nick gabbler we have collected a series of feed int uh data by pen during the first week especially uh this uh looks very messy but is data on Feed in take by pen during the first seven days so multiple trials so every line will be uh data for a pen and we're adding more data and what we want to do is to understand and how this early Feit in take is a predictor of the body weight at day 42 for the peg so that's one of the main reasons and this is why this is this is uh taking that data grouping the data into low and high fitting take pics uh the high fitting take pics being in a in a black uh in a black color and the low fitting take pics being in a red and you can see that this becomes the early fitting takes becomes uh an end of the nurer good predictor of performance for average daily gain and for fit in take in this graph below is showing each of the days how that fitting take um is being affected through the period of the nursery so that's our interest into understanding Feit in take on the first three days and applying a computer vision technology now I'll moving to another of this SM Farming tools that we are working and is the digital is it involves the digital transformation now not going to go through every of these steps because G get gave a really good example of what that means bringing the data into a cloud integrating but the most important piece here is how you activate the data right now right now I think is our goal uh from The Faculty from the university is to put in in the students hands and to in training them how to use these new technologies that are becoming more available to understand the data that's been integrated so um Nathan came with the passion for data analytics and the passion for pic and so some of the questions that we're working are around the accurate marketing and sales of pics I believe there's three key components to be accurate into this area and is I need to time right the first sell of the pics um I need to because that's going to drive my subsequent sellings but I need to also be precise in selecting the pigs that I need right so if I am able to track how pigs grow over time um you might not see at the back it's it's small again but it the computer the algorith uh maybe the the camera can tell me how pics are growing and how that variation is changing over time why is that important it's because if I can do an accurate marketing process I can avoid something like this so this is a producers data of 20 22 what I'm showing here is the only the first cut aage weight picks the target for the producer is 285 so I want to sell by first cut of PS at 285 pounds I got very accurate at 287 and that's using tradition and me methods right I want to understand how PS grow so I apply grow CPS I apply equations and estimations and I do get very accurate but uh I only get 62% of the p within 10 pounds of that Target meaning I'm not precise into into the timing of that first cut and that's what I meant when I said when I said that there's about three to$ two to five dollar per pick opportunity in in the right um process of marketing that's where Nathan is having a Passion he start working on this is an example of what that what I meant with uh the big opportunity we have and maybe the application of Technologies this is in collaboration with a a uh Pig Vision company where H one of the bars here shown into um into the reddish color Barn one follow a standard marketing uh protocol which was Guided by grows and fit curs uh so again accurate but not precise and then we challenged that with a barn two that was Guided by the computer vision and AI so the system will tell how pics are growing over time and when went to market the first cut of PS and went to market the subsequent pics now I'm comparing these three um these three bars that are the times and the percentage of pics that I sold my pics from the standard procedure and the four bars in the darker color are what the camera and the AI will say pics need to go now comparing here here in the table the weights are very similar is for the standard protocol it was two about 290 lbs versus 29 289 pounds so no gains into the body weight of the pigs but notice that I save a week so there's a six days Less on days on feet for this group of pigs what that mean is um if I look into the distribution of those carcass the standard protocol will be a lot more variable that my computer vision and AI computer vision and the right timing means I want most of my picks in the right uh weight ranges right because that's what's going to optimize the price that I get paid by the plan uh processor right if I go too high or too low in my weights when I sell my pigs there's great penalties that I can avoid if I if I do a more accurate and precise marketing protocol now how we measure there was some changes in the body way uh but the variation were was reduced by 2.3% in this case when I compared the two so that's the opportunity I'm talking about you're just optimizing the number of PS that go into the right ranges of Weights um as an example of the application Nathan he started with us last semester as a we as a sci with practice um program and he start developing some of this modeling so he develop a grow curb and along with that grow C curb he developed a peak marketing tool and so here's just an example of tight trading what he did is if I take my pics at 140 to 180 pound uh 180 days right and I have changes changes uh changing conditions in my market so I have feed costs that are higher or lower and my pick price that can may be higher and lower that was um given to producers that then they can they can change those factors um so this is very early on too but it's a very nice example for him and for his science with practice project he got award an award for uh most impact to the community when he presented this project at the University so uh this is the area where he will be working uh and expanding so expect a lot more from Nathan on this area and last topic I want to talk is is very close to my heart is the area where I can play with sa and work with SA but also bring to the industry anir a a nutritional intervention that I think it it will have a lot value but there's a lot a lot of work we have to do in this area I just want to highlight some of the value that dietary fiber has in the different phases of the of the cell just anywhere from gestation uh peripartum lactation improving the satiety reducing inflammation reducing constipation which has been proving reducing improving the colostrum production improving the feed intake so I think there's benefits unfortunately in the US I think we don't have as much access to fibrous ingredients for Sals as s countries like Europe and two uh we need a better understanding of those components so here is a a a literature review uh of ten studies it was provided by Dr pit wilker to us and it's just to illustrate uh the difference between a control uh on those studies that had 300 gram of total dietary fiber as a measure of the fiber and a high fiber which was more than 500 gram per day and this is during gestation uh calculated in a grams per day basis and so uh in the review what he notice is that there is an improvement in the number of pigs win and the leader weight gain and the feed intake so some of those key aspects is what we want to bring it to to our type of diets uh to understand more of our ingredients and how to get there but it's not only the level of total dietary fiber it's also the composition of that fiber and is in this case is showing the response to increasing levels of soluble fiber from 60 to 90 meaning that if you go close to 70 um close to 76 or above close to 70 grams per day of soluble fiber there will be a response to number of PS win and the lead their weight gain so that's encouraging we're taking and we took in our conversations uh that encourage us to try it into the field and and specifically targeting uh sou mortality so we what we did here is to take a South Farm that was a high mortality Farm it was highly productive but also High mortality and we want to challenge ourselves with our levels we believe we had high levels of total dietary fiber because of our of our access to wheat millings in that area um so we were about 13 14% total dietary fiber with those 30% uh weat mealings uh but because you have to challenge yourselves sometimes we went to 18 in a first phase but then push farther to a 30 more than 30% total dietary fiber and to get there um we had to do uh 20% whe mealings and 40% soybean holes because we wanted the right level uh of fiber but also the typ of fiber and the Sol fiber in this case will be um bringing from the soy holes right we don't we don't have much access to soluable fibers uh but the one that can give you some will be soy holes if you want to get there we use 610 sou we did this in the field and again a highly a high mortality South arm uh so this is what it encouraged me to continue this work and to pursue more in this area when we measure mortality that's the area we were more interested about we saw a a tendency to reduce the mortality relative to the number of ss that we placed in the trial when we fed the high fiber diet now specifically and this is a pen station Farm we saw that the mortality was mainly driven by deless reduction into that farm uh but relative to the discussion we having in the industry to prolapses there was a numeric reduction as well so no as drastic as we we wanted to see it's more of the impact in lameness to the cells but very encouraging data um now if you go by regions they you it will be very difficult for us to get there right and that's what we need to understand our ingredients just just to give you an example here a corn soy diet it will be about 8% total dietary fiber so if you put in a grams per day basis that will only provide about 180 no more than 200 grams per day right and when we compare to European data U they will be feeding about 450 to 500 grams per day now you might say you have access to ddgs and you can put that into your regation diet you might get to a 10% a little more on total dietary fiber uh and that doesn't get you any closer to the 450 grams per day right um compared to a European diet so We compare these by different regions and different diets these are Brazilian diets Spain UK T Thailand and us this is just to give you an example of where we are in the US relative to Europe specifically right if we want to chieve 450 grams per day we're very far from it too and spe especially if we go and look the soluble fiber which we said might want you might want to be might want to be between 60 and 70 gram per day we are half there right because we don't have the right resources um now I believe these are the research needs I mentioned to you earlier these are some of the work we're already doing especially into characterizing the fiber into um the US ingredients so that's great work we're doing with Dr Petri but then we need to understand what's the contribution of those fibers ingredients to the energy value for the sou specifically which I don't think we have a could handle of what's the functional value of that dietary fiber but then two is as we bring that fiber imagine you have to do 40% soy holes into your diet that's going to change the Dynamics and the feed Logistics right is you have to accommodate uh load more volume because it's less dense into the fit meal your diets are going to change the density if you don't if you don't pellet so some of those will be practical constraints but if we see the value of the fiber in reducing specifically mortality and we heard from Dr uh from um Steve this morning that mortality is one of the key limitations into our industry and challenges so I think the value is there the needs are there that's where our group is working on and lastly I I just want to acknowledge a group of people that in this first year I have started to collaborate they have been very open to establish this conversation and and started these projects uh with us Dr Emy Petri Nick gobler Dr Griner have been a great nutrition group here at dawa State and now into the computer vision Juan Santos pande um Anna Johnson is bringing a great component into the animal behavior so I'm I'm very delighted to have here uh and collaborating and then a group from the BM Dr Danielle lares Gustavo Sila now we have a Dr Edison mag mahalis now joining uh our department so with that um like to appreciate your time thank you for staying to the end it's the last session um I'm honored to be here um in Iowa State and it's been a pleasure thank you not yet we want to make it available that's that's the goal and we'll make it available through iPic there's a great Dynamics there and so you you can access to it yeah yeah certainly it's an area where there's a l Of Interest right I think it's the the settings we have to understand mortality relative to nutrition are are no ideal you need large data sets and in a control way we try to do some something like that in the sou and and also we have to push the limits and if if you see the example I provided We compare um 133% to 30% which is huge and and you don't see those type of diets being compared but I think it's needed if you want to see something like mortality so probably you have to go beyond what you know and then two you have to have the right setting so you have to have a sound number of experimental units that that can answer mortality questions but but to your point yes that's of our interest as nutritionist if I seen a couple of examples silen a would be another one where there was field experience we also try to measure into the field into the right conditions but uh targeting our metric to be mortality yep that's what I will assume with fiber uh John um especially being a pen station is and we didn't measure uh we couldn't um that's our purpose of starting to develop some of the computer vision Technologies it will get us there uh but some of the impressions in the field where the pigs sou will were more calm after eating I think that reduce Locomotion that reduce walking and behavior issues probably okay thank you very much [Applause]
2024-08-20