NZ615942B2 - Apparatus and method for using infrared thermography and behaviour information for identification of biologically important states in animals - Google Patents

Abstract

Disclosed is an apparatus for identifying biological states in an animal. The apparatus comprises means for identifying the animal (6) to obtain animal identification information, at least one infrared thermography camera (8) for photographing the animal to obtain infrared thermography information and fidget information about the animal and a processor (9) for receiving and processing the animal identification information. The infrared thermography information and the fidget information can be used to determine a biological state of the animal. The biological state is a predictor of onset of disease, growth states and reproductive states in the animal. nd fidget information about the animal and a processor (9) for receiving and processing the animal identification information. The infrared thermography information and the fidget information can be used to determine a biological state of the animal. The biological state is a predictor of onset of disease, growth states and reproductive states in the animal.

Description

TITLE: Apparatus and Method for using ed thermography and behaviour information for fication of bioiogically important states in s TECHNICAL FIELD A non—invasive apparatus and method of fying biologically important states in animals is provided. More specifically, an tus and method of combining real-time, non-invasive, ethological behavioral information with infrared scanning is provided for identifying agriculturally important states, such as disease, growth, or reproductive states, in animals such as livestock.

BACKGROUND Livestock often undergo significant exposure to transport and handling, co—mingling, n and some time off feed and water. tively, these events can impede the immune system and can result in a significant incidence of disease. Such events can have considerable economic impact, for example, on the agricultural industry both with respect to health treatment costs and animal performance. Recent research has resulted in an increased understanding of the importance of animal management factors, such as transport and ante—mortem handling, in influencing both animal welfare and the food quality arising from such animals. It is known that disease and stress can have a dramatic, negative impact on animal welfare parameters and performance as well as meat quality and yield and hence the economics of the animal industries.

As would be known to a person of skill in the art, a number of diseases, such as Bovine Viral Diarrhea (BVD) type 1 and 2, infectious Bovine Rhinotracheitis (lBR), Corona Virus, Bovine Para—influenza (PIB) and Bovine Respiratory Syncytial Virus (BRSV), can impact iivestock populations. One such disease complex, known as bovine respiratory disease (BRD), refers to a host of complex diseases, and is lly used to refer to an animal ying an undifferentiated fever and/or other clinical signs (eg. respiratory distress, lethargy, and loss of appetite).

The presence of BRD in intensively raised calves has caused a dependence on antibiotic treatments (including mass treatments), which, in turn, has led to a n for the promotion of antibiotic resistant microbes. indeed, the ability to treat BRD in cattle is becoming more difficult due to the emergence of resistant microbes (for e.g., pneumonia), or new zoontic diseases in multiple sourced, co—mingled cattle. rmore, recent reports have shown substantiai contamination of s and meat products with antibiotic resistant strains of bacteria such as E. coli.

The effectiveness of treating livestock diseases, such as BRD, can depend upon the ability to detect, diagnose and treat affected s early. The ability to achieve early detection will depend upon the information available and on the reliability of that information. For instance, when used alone, traditional al signs of e provide poor diagnostic results because clinical symptoms often occur late into the course of the illness. Further, many diagnostic techniques, such as the use of acute phase proteins or hematology assessment, require the capture and ve, in vivo collection of biological samples, which result in the significant cost of analysis and time. The requirement of the capture (and therefore restraint) of the animal in order to collect a ical sample causes stress, and the process itself is therefore introducing inaccuracies into the data collected.

Recent research has focused on alternative approaches to non—invasively determine the early fication and onset of disease in cattle. One such approach is infrared thermography, which can be used as a means of detecting the dissipation of heat in s. Thermography operates on the principle that infrared radiation can be utilized to observe radiated heat loss and to provide an early indicator of fever because up to ~60°/o of the heat loss from an animal can occur in ed ranges. The technology has been demonstrated to be effective in non—invasive identification of transport and other environmental stressors that can alter an animal’s heat loss. r approach to non—invasive disease analysis is the commonly used “pen-checking” approach, wherein the animal caregiver observes the animal on a daily basis to detect any al behavioural patterns, or clinical signs of illness (e.g. decrease in eating clue to loss of appetite; see Table i for further examples of behavioural arks). Although non-invasive, pen-checking is highly inaccurate particularly during the early stages of disease onset and leads to false—positive and false—negative results because it depends upon the skill and observations of the ver. Further, it is known that animals often do not display overt signs of illness (that would be detectable to a ver) until later in 1O the progression of the disease, resulting in an increased risk of infection of healthy animals in a population, particularly where the animals share a source of food and water.

TABLE 1 — PRIOR ART Commonly—used clinical scores used in the bovine respiratory disease (BRD) early disease ion , Clinical Score Assessment 0 4 5 Disposition, Moving Slightly Moderate Hanging Prostrate, Death Lethargy around depressed lethargy back from ent and well with appearance, and the rest of or abnormal Behaviour normal Holds head depression, the herd, posture, e, slightly Holds head Recumbent Not Content, lower than low, Droopy or interested in No normal, Mild ears, Slow abnormal surroundings, signs of anorexia to rise, posture, Weakness lethargy Stiff Largely movements, depressed Anorectic Respiratory Normal Very fine Fine crackle Medium Course Marked Insult breath crackle and/or crackle crackles respiratory sounds and/or moderate and/or and/or distress moderate nasal moderate severe and/or lung cough discharge to severe discharge idation and viscous with moderate nasal atory cough discharge distress and with cough obtunded lung sounds Digestive E No Mild or Moderate Moderate Severe Severe Insult E insult, slight diarrhea to severe diarrhea, and diarrhea and 1 Normal ea with 10% diarrhea E less than 1 not eating, i l ‘ eating with slight dehydration with 10% or [ 10% of not drinking ; and dehydration and less of feed t normal feed and l drinking and reduced reduced intake and ' intake 1 ated 1 eating feed intake more than i 10% i 1 dehydration l It is also known that the identification of non—disease states in animals is important to the agricultural industry as well as to 200 and wildlife y settings. There are many biological events in an animal’s life that influence a plethora of biometric measurements and characteristics expressed. Some of these events are normal biological functions an animal will display such as when they adapt to a ng environmental ature, a changing growth period or a changing endocrine event ing puberty or estrus. Other events are less common and will include the onset of a disease state. In either disease or non~ 1O disease states the animal will be ered to be in a biologically important, non-steady state during these periods. These biologically important states may have, for example, agriculturally important consequences and implications.

Growth ency in animals is often defined as the gain in a particular tissue type such as muscle or milk compared to the input of resources such as feed and water. ln addition to disease states, growth efficiency is an important attribute in animal agriculture as competition for limited resources increases.

However, measuring growth efficiency has always been a challenge. One of the more accurate methods to monitor growth ency is to use indirect metry which measures exactly the amount of oxygen and energy used by an animal for a given increase in gain of a specific tissue while noting that the metabolism will also give off heat (Kleiber, M. 1961. The fire of life — an introduction to animal energetics. John Wiley & Sons, inc). Alternatively, growth efficiency can be monitored by measuring the actual feed consumed by an animal and the growth that resulted or measuring the so called gain to feed ratio (Kleiber, M. 1961. The fire of life - an introduction to animal energetics. John Wiley & Sons, Inc).

A more recent approach to monitoring growth efficiency has been to monitor the so called residual feed intake (RF!) which fundamentally is a comparison of the ed feed to gain against a known estimate for feed to gain based on scientifically accepted formulas (Basarab et ai. 2003, 2007 see below) while a . However, this later method, reasonably accurate, requires lengthy seventy days or more feed monitoring period which is both expensive and impractical.

It is also known that the identification of uctive states in animals is important to biology in general and to the agricultural industry specifically. For example, reproductive states such as onset of puberty and estrus are important to identify for the purposes of reproductive efficiency, and ore agricultural efficiency. it is known in the art that the onset of puberty and estrus are characterised by behavioural estrus which includes an increased ssness of the animal.

There is therefore a need for vasive, early and te means of identifying biologically important states in animals. Furthermore, there is a need for a non-invasive detection means that are capable of identifying diseased animals, even in populations where there is a low prevalence of the disease.

§QMMABX The present apparatus and method provides for real-time automated, non— invasive infrared thermography information of an animal to be used for both thermal and oural measurement, thereby providing an r and more accurate tor of onset of disease, growth states, or reproductive states in that animal. More ically, the present system and method provide for the use of thermal images (taken, for example, at a water station) to obtain both temperature and behavioural information about one or more animals at a time, and to utilize that information to determine the health, growth, or reproductive state of the animal. The combination of thermal biometric data, such as radio frequency identification ed thermography, and behavioural biometric information, such as behavioural fidgets can be used to detect early-onset of these biological steady and non-steady states in animals.

Broadly speaking, an apparatus for identifying important biological states in an animal is provided, the apparatus comprising: an enclosure for receiving the animal therein; means for animal identification mounted on the enclosure and connected to a reader for identifying when an animal is received into the enclosure; at least one infrared thermography camera mounted on the enclosure for photographing the animal to obtain infrared thermography and behavioural information from the animal; and a processor in communication with the reader and camera for receiving and processing information from the camera and the reader; wherein the information processed by the processor identifies important biological states in the animal.

Broadly speaking, a method of fying ant biological states in an animal is provided, the method comprising: providing an enclosure receiving the animal therein; receiving an animal within the enclosure; identifying the animal; raphing the animal to obtain infrared thermography images and behavioural information from the animal; processing the infrared thermography images and behavioural information; and identifying ant ical states in the animal as a result of processing the ed thermography images and behavioural ation.

Definitions of specific embodiments of the invention as claimed herein follow.

According to a first embodiment of the ion, there is provided an tus for identifying biological states in an animal, the apparatus comprising: means for identifying the animal to obtain animal identification information; at least one infrared thermography camera for photographing the animal to obtain infrared graphy information and fidget information about the animal; and a processor for receiving and processing the animal identification information, the infrared thermography information and the fidget information to ine a biological state of the animal, wherein the biological state is a predictor of onset of e, growth states and reproductive states in the animal.

According to a second embodiment of the invention, there is provided a method of fying biological states in an animal, the method comprising: identifying the animal; photographing the animal to obtain infrared thermography images and fidget information from the animal; and processing the infrared thermography images and the fidget ation to identify a biological state of the animal identified, wherein the ical state is a predictor of onset of disease, growth states and reproductive states in the animal.

All documents and references referred to herein are incorporated by reference in their entirety.

FIGURES Figure 1 depicts an embodiment of an apparatus for combining real-time, non-invasive gical behavioral information with infrared scanning for fying biologically important states; Figure 2 shows a schematic diagram of the embodiment of Figure 1, image as published in Schaefer et al. 2011. Research in Veterinary Science. In Press; [Text continued on page 7] Figure 3 shows a graphical representation of our data for calves suffering from BRD, wherein the ill (sick) calves “fidget” more per drinking bout ed with healthy (not sick) calves; Figure 4 shows a graphical representation of behaviour data for calves suffering from BRD, wherein the ill (sick) caives “fidget” more overall than healthy (not sick) calves; Figure 5 shows a graphical representation of infrared thermography data vs time of true positive (ill) and true negative (healthy) calves with BRD; Figure 6 shows a graphical representation of infrared thermography data d against time for true positive (ill) and true negative (healthy) calves with BRD; Figure 7 shows a graphical representation of a ison of the infrared thermal values in a True Negative (TN) healthy animal and a True Positive (TP) ill animal for ison. Data shows the ed thermal value (y axis) vs the day of study (x axis).

DESCRIPTiON OF MENTS An apparatus and method of eariy detection of biologically important states in animals, such as livestock, is described. In some embodiments, the biologically important states can be agriculturally important states.

More specifically, an apparatus and method of combining thermal and behavioural biometric information for the early identification of disease, growth efficiency, puberty, or estrus is provided. infrared thermography (lRT) and ethological benchmarks may be combined to provide an early and automated identification system. While the present disclosure generally s to beef cattle, it would be understood by one skilled in the art that the apparatus and methods provided herein may be utilized to detect disease in any , such as livestock species, including, but not limited to dairy cattle, pigs, and poultry.

In the present apparatus and method, automated thermal and ethological data may be ted simultaneously using at feast one lRT camera and software system. Thermal data can be used in conjunction with predictive or diagnostic infrared values, alongside ethoiogical (behavioural) predictors termed "fidgets" (Le. the "fidget factor”), wherein both the ed values and fidgets are both determined from the lRT thermal camera image data.

Infrared Thermography ation An embodiment of the present scanning apparatus is shown in Figure 1.

Figure 1 depicts a an embodiment of an automated, radio—frequency identification—driven , multi-cait infrared scanning apparatus which can be attached to a water trough, and can comprise a data storage unit (A), camera housing (B) and a water system with an RFlD antenna (C).

Having regard to Figure 2, the automated scanning apparatus can comprise: 0 An enclosure, for receiving the animal therein, n the enclosure can be, for example, a water or food station. In some embodiments, the enclosure can be fenced in, in other embodiments, the enclosure could simply be a pasture area. As would be understood by one skilled in the art, an ure can comprise any area or structure which accomplishes the functions described herein.

In the present embodiment, a water station was designed having two side panels (1), surrounding a commercially-available, two—water bowl float design (2) from Ritchie water systems (Ritchie Cattle Fountains, Conrad lA, USA), and optionally ted by a partition there between. The present ure may allow for access to the water station from two or more directions. it should be noted that a water station was utilized in the present embodiment because during illness it is known that s cease , due to loss of appetite, before they cease drinking. lt shouid also be noted that any enclosure, or other means of reducing overall movement of an animal, in a —free way, such that an image could be taken of the animal, could be utilized and is contemplated; Extension panels (3) can be piaced on each side of the water bowls to “centre” or frame the position animal’s head, and to help keep the animal’s head at the proper focal distance. A panel (1) on one side of the water bowls may be modified to facilitate a window (4) in order to View the animal while at the water station. The window may measure, for example, approximately 30 cm square; At least two in-phase loop antennae (5), for receiving information from the ’s radio~frequency identification tags (RFID), or other such animal identification means, as applicable, may be mounted in the panels (1) adjacent to (or near) the water bowls (2), and ted to an Allfex PNL- OEM-MODLE-3 RFID control module or “reader” (6) (Allflex ElD system, Allflex Canada lnc. St—Hyacinthe, P.Q.).; At least one infrared thermography camera (8). The camera (8) may, for example, be capable of obtaining at least 1 — 60 images/second such as a FLIR 860 broadband camera (FLlR Comp, Boston, MA), which may be rotably mounted adjacent to or near the windows (4). Means for electronically rotating the camera (8), such as a geared—head motor, may be connected to the camera, for powering the rotation of the camera. The camera may be used to obtain radiated temperatures around, for example, the l area (eye plus one centimeter nding the eye) of animals. The orbital eye area was chosen in the present system because it is known to provide an accurate peripheral temperature reading, thereby providing a measurement that is sensitive to both stress and disease onset. Although the thermal orbital (eye) is described herein, it is understood that any area on the animal that provides an accurate and adequate peripheral thermal g of the animal’s temperature may be used; and A l system or processor (9), for receiving and processing information from the camera (8) and the RFID antenna/reader (5, 6). For ce, the processor may be mmed to control the camera positioning, acquire the infrared image, perform the anaiysis of the image data, and to store the acquired information on a database. The information may be collected and received tically upon the animal entering the enclosure, and the processor may collect the ation via wireless transmission, such that information may be monitored remotely. ment integration, and the re and software used in such a thermal station was designed and developed, in part, at the Lacombe Research Centre, Lacombe, a, Canada.

In one ment, an optional electromagnetic shielding (7) may be exposed to the g pen on the side of the panels (1) to prevent the improper reading of RFiD tags on s that are not within the enclosure. in operation, the present apparatus and method provide that when an animal enters the enclosure, the RFiD antenna system (5, 6) can receive the identification of the animal from the RFlD tags, and can signal control system (9) to rotate the camera (8), if necessary, in the direction of the animal, and to initiate capturing images of the animal’s head when it becomes visible through the window (4) in the panel (1). in one embodiment, the camera/motor assembly (8), can be enclosed under a protective cover, and can be located medially between the two viewing windows (4) at a distance that provided a field of view to cover most head positions of an animal. it is understood that mounting the infrared camera on a motor capable of rotating to at least two different scan windows (4) as signalled by the RFiD reader can provide the capacity to obtain information from at least two animals at one time. For instance, the system may be ed to accommodate a second enclosure/thermography station ed parallel to the first station with the camera located centrally between the two ns, thereby at least doubling the animal handling capabilities of the system.

The present apparatus and design may enable known methods of correct thermography ques, namely, a fixed focal length and angle with a near—still image, thereby providing accurate thermal data collection. The system can further provide non-invasive means of obtaining both thermal and behavioural (discussed below) biometric information without the need to restrict or capture the animals. it is understood that any similar system capable of ing t thermography information, without necessitating capture and restraint of the , is contemplated.

Ethologica/ (Behavioural) Information The present apparatus and method can also provide for the use of the lRT information obtained from animals as a measurement of behavioural prediction of disease onset or other biologically important states in animals. For instance, the ’10 thermal images taken at the water station, as herein described, can be further utilized to obtain behavioural information about the animal, thereby providing means for combining the thermal (temperature) biometric information with behavioural biometric information, to provide earlier and more accurate disease ion and state identification in the animal.

Each tamped image taken by the automated lRT system can be classified as a behavioural “event”. For example, movement of the animal at the water station can result in the camera (8) having to reset the thermal contrasts between and among the lRT pixeis, thereby automatically causing a new image to be taken and time—stamped and the postural adjustment or “fidget” of the animal to be recorded. As such, depending on how much fidgeting the animal does at the water station, more or fewer images may be taken of one animal as compared to r. The images can then be used to calculate behavioural factors such as, for e, the total time the animal spent drinking, the number of drinking bouts, the length of each drinking bout, the average number of drinking bouts per day, and the number of “events”(i.e. fidgets) recorded (eg. the number of thermal scans taken during a single drinking bout). The ation can then be used, in conjunction with the thermal information to determine “true- positive” (i.e. sick) and egative (i.e. non—sick) disease. ingly, an ethological tor of disease, referred to herein as a “Fidget Factor” can e an additional benchmark for non-invasive disease detection and state identification.

It should be noted that different animals may fidget more or less than others in the population, and that fidgeting can r be altered due to an illness, growth state, or uctive state. it should be known that the processor can be capable of utilizing all of the infrared thermography images of each animal (eg. orbital (ocular), mouth, nose, ear, shoulder, and body images were all included in the behaviour data set) in order to process the fidget our.

Thus, the t apparatus and method can provide for the use of infrared thermography images to be used to detect the peripheral temperature of the animal as well as the behavioural activity of the same animal, thereby providing earlier and more accurate disease ion and state identification. it is understood that the present apparatus and method can provide for two distinct sets of data or information to be generated in parallel or series. it would also be apparent that these two biometric data sets consisting of both infrared and fidget information can be used in a number of statistical assessment procedures including multiple regression and correlation, ranking and prediction indexes to enable the more accurate identification of true—positive and true negative Such detection and identification means are likely to be applicable in a variety of settings, including, for example, in bio-security and bio—surveillance circumstances.

The ing examples are provided to aid the understanding of the present disclosure, the true scope of which is set forth in the claims. it is understood that modifications can be made in the system and methods set forth without departing from the spirit or scope of the same, as defined herein.

EXAMPLES Example 1 Animals In this example, forty (40) multiple sourced, co-mingied and transported commercial, auctioned er calves, which had been exposed to viral and bacterial infection for respiratory viruses including BVD, Pl3, lBR, Corona and BRSV, and forty (40) retained possession calves were used. The calves were weighed, monitored for core and orbital thermal properties, blood sampled and placed onto conventional cereal grain silage with access to shelter and clean water.

Twenty (20) of these calves were obtained from the BVD and lBR antibody free herd at the Animal Diseases Research lnstitute at Lethbridge, Alberta, Canada. These calves were Angus x Hereford crosses and had been weaned approximately one week prior to transport to e, Alberta, Canada.

The calves were an average of 550 tbs, were raised on native grass pasture and had been given a de-worming medication two weeks prior to weaning.

The calves were transported on a tional ected horse trailer.

On arrival at Lacombe Research Station, a ort time of approximately 5 h, the calves were co—mingled with 20 muitiple sourced and commingled auctioned calves. All calves were monitored continuously for 3 weeks. All calves were observed to have continuous contact with each other by touching noses as well as sharing the same water trough, salt lick, feed bunks and g.

Infrared Thermography Automatic ed thermography images (lRT) were collected using a portable lnframetrics broadband S60 ed scanner (FLIR® lnframetrics S60, Boston, MA, USA). All images were taken of animals as they entered the automated infrared ng station at the common water station located in the pen (see Figures 1A, B and C). in the present experiment, images specific to the orbital area of each calf were used in collecting thermal data.

All calves were thus monitored for average daily maximum temperatures (including change in temperature) and for the mean ratio values (MR). The mean ratio which was calculated as the average of daily radiated maximum temperature for a given animal divided by the average daily maximum value for the group of calves. The thermal data was verified by ison to serology and virology blood parameters.

Ethology — “Fidgets” Using the time stamp (which provides the hour, minute, second and date of the image obtained via a chronometer or clock), for each image taken by the automated lRT system, each image was defined as a behavioural “event” which triggered the image to be recorded. During the present example, a 4—minute interval between ng events (known as the bout criterion interval; BCl) was used to determine the termination or conclusion of one ng bout and the start of another. Accordingly, the same infrared images were used in the analysis of both thermal and ethological data sets, albeit, the ethological data set ed “all images”, while the thermography data set included “orbital images” only.

Results in the forty (40) multiple—sourced and co—mingled commercial calves, there were 10 s out of the 40 fied as “ill” (true—positive animals). This identification was made by virtue of displaying clinical scores of 3 or higher (see Table 2), and by orbital infrared values of 351°C compared to healthy animals (or true negative animals) having a ature of 348°C. Clinical illness was verified by statistically significant haematology values. in addition, the ill calves demonstrated an approximately 40% increase in the blood cortisol values (deviating from an average of 52 nmol/L in healthy calves to over 70 nmol/L in ill calves). Results demonstrate that haematology data for the forty (40) l retained possession calves displayed normal haematology values.

It is known that four to six days prior to the display of clinical signs and lab verification of illness, infrared orbitai scans can be 71% efficient (combined true ve and true negative ) at early fying ill s compared to either clinical scores alone (55% efficiency) or rectal temperatures alone (59% efficiency). This is supported in the present example where the vasive collection of orbital infrared atures, alone, was 73% efficient at identifying ill animals 27 days before clinical symptoms able by “pen-checking”.

Further analysis of all thermal images recorded through the automated lRT system, and based on a 4 minute drinking bout interval, results show that true—positive or “sick” animals have a tendency to “fidget” more than “non-sick”, true-negative animals (Figures 4 and 5). Sick (true—positive) animals were found to have a greater number of lRT images taken during each ng bout. This is 1O despite overall ng behaviour, which includes the drinking duration and the number of drinking bouts, being the same in sick and healthy (true—negative) animals. Based on the number of average fidgets (events) per bout in sick and healthy animals, a behavioural predictor or “Fidget Factor” of 4 fidgets per drinking bout was determined to be a possible indicator of disease.

False-negative and false—positive animals were not included in this data set, as analysis focused on true-sick and true-healthy animals only.

TABLE 2 Means iSD of haematology values for the forty multiple d co—mingled calves. White blood cells (WBC) and all other ential cells = cells X 10 Red blood cells (RBC) = cells X 10 12, hgb = g/L WBC Neut Lymph Mono 1 Eosyn Baso RBC HgB Hct% N/L j L Healthy 9.12 1.22 5.56 0.98 I 1.28 0.07 _8.83 11.98 35.4 0.24 SD 1.65 0.83 1.15 0.51 ‘ 0.87 0.07 0.94 1.08 3.8 0.22 (n=30) HI 1239 3.33 5.24 2.11 2.14 I 0.13 7.94 11.24 34.4 082* tBRDl .

SD 3.78 3.23 2.26 1.62 1.87 0.01 1.04 0.91 3.6 0.91 (n=10) P value 0.01 0.01 0.55 0.01 0.01 0.02 0.01 0.04 0.41 0.01 Statistical separation based on least squares analysis (two tailed t—test). * N/L ratio for ill animals was either very high or very low.

Haematology, Endocrine and Serology Data With respect to laboratory analysis, salivary and serum cortisol was ed using a known enzymatic assay from ted samples. Hematology analysis and differential counts were conducted on a Ceil—Dyne model 3700 hematology analyser t LabsTM, Mississauga, Ontario). Serology assessment was conducted by Prairie Diagnostic ServicesTM (Saskatoon Saskatchewan) and assessment was carried out for the BRD s, Bovine Viral Diarrhea (BVD) type 1 and 2, as well as infectious Bovine Rhinotracheitis (iBR) via serum neutralization tests.

Additional assessment for Corona virus, Bovine Para-influenza (PIS) and Bovine Respiratory Syncytial Virus (BRSV) were conducted by ELlSA using methods known to one skilled in the art. Antibody trations (units) for BVD, iBR, BRSV, Pi3 and Corona were obtained as follows: ((mean net optical density of sample — mean net optical density of fetal bovine serum) / (mean net optical density of positive standard —- mean net optical density of fetal bovine serum)) X 100 The g of antibody titre scores was as follows: for BVD and iBR 0—2 = negative, 3-1321 = suspicious, 14-40:1 = low, 41~80:1 2 moderate, >80:1 = high.

For BRSV, PB and Corona <10 = negative, 11-13 = suspicious, 14-50 3 low, 51:100 = moderate, > 100 = high.

Example 2 Animals in this example, r trials were conducted on 100 le sourced, co— mingied, transported and weaned commercial calves. These calves were procured from two primary sources with 17 from the ADRi herd at idge, Alberta, Canada and 83 procured from commercial auction facilities. These calves were themselves purchased through auction from two separate locations.

The calves were brought to the Lacombe Research Centre (LRC) Beef Research unit, weighed, blood sampled, core temperatures recorded and then placed into clean receiver pens with wood shavings bedding and free access to water and cereal grain silage.

Methods Continuous, tic ed and oural data was captured on all animals for a three—week period (except when power supply or solar loading glitches caused a failure in the system), as described in Example 1.

Results In the present example, thirty seven animals were identified as “at risk” of BRD by the “pen checking" technique. Of these animals, 24 were subsequently verified by objective lab data as being true positive (TP) and 13 were identified as false positive. Hence, the incidence of false negatives and positives is again comparatively high when pen checking ciinical scores alone are used to identify BRD. in the calves identified as true positive for BRD, the RT temperatures were on average 367°C compared to the true negative animals at the same time at 356°C (P<0.05). Figures 6 and 7 demonstrate the relationship between true- positive and true—negative animals for calves with the most complete clinical score and infrared data, and show that the infrared scores detected animals that were verified to display BRD several days before the ciinicai “pen checking” scores.

Using the “Fidget Factor” tor generated and d in e 1 to detect sick animals (an average of 4 fidgets per drinking bout over a 24 hour period), no significant differences were found in the present Example 2. Some possible expianations may e either the need to alter the drinking bout interval or due to the high rate of false negatives in caif group. The present example also experienced higher than usual electrical glitches, which may have made the gical data set less robust. t research on adjusting the BC! and the fidget predictor value is required and continues.

Example 3 Animals in the present Example 3, investigations were d out on a total of sixty—five (65) receiver calves. These calves consisted of 54 retained possession, low disease incidence animals from the Lacombe Research Centre (LRC) Beef Research fall calving herd and a further eleven (11) calves from the high-health, closed herd located at the Animal Disease Research Institute (ADRl) at Lethbridge, Alberta, Canada. The ADRI calves were unique in that they displayed no antibodies to either BVD or IBR virus and therefore would be susceptible to BRD-causing viruses. Calves from both herds were commercial crossbred cattle of British X Continental breed. All calves had been weaned and transported to the LRC beef unit prior to the study. The animals averaged 220 kg at the start of the study.

Methods To simulate l marketing conditions all calves were co—mingled and transported to a local auction facility within one hour of the LRC beef unit. The calves were then offloaded and kept in pens overnight without feed or water. The cattle were loaded onto a cial r and returned to Lacombe, Alberta, Canada the following day for processing.

On arrival at the LRC beef unit the calves were weighed, blood sampled, core temperature recorded, clinicaily scored and then placed into receiver pens containing straw bedding and with free access to water and a cereal grain silage.

The calves were subsequently “pen checked” daily for signs of s and lRT values were recorded continuously using the present system and method as defined herein. Behavioural “events" and/or s were monitored and an alternative ”Fidget Factor” for disease was determined based on various intervals between drinking bouts (eg. 3 or 5 minutes, rather than 4 minutes utilized in Example 1). in addition, preliminary live observations were conducted to determine which specific faciai nts ed IRT images to be ed as events (i.e. to define the specific mechanics of a “fidget”). it is contemplated that such ations could be further enhanced by classifying and defining the actual fidget our by, for exampie, video analysis. logy values for all animals were assessed on a CellDynTM hematology analyser. Clinical scores were assessed using a point system (see Figure 2) and core or rectai temperatures were recorded using a chute side digital temperature probe.

Results Based on the pen checking information, two of the sixty-five calves were considered to be at risk of BRD. Taken together with data available at the time of 1O processing (core temperatures of 40°C or higher) four of the sixty—five animals would have been diagnosed as true positive for BRD. However, two of these animals were subsequently determined to be false positive by hematology is (white blood cell numbers and neutrophile/lymphocyte ratios), core temperature and clinical score. The use of the same analysis would also have classed eleven (11) of the calves as true positive (TP) and 20 as true negative (TN) with the remaining 34 as intermediate health. Three of the true ve (TP) animals were from the ADRl BVD and IBR antibody free herd and 8 from the LRC herd. The average values for these animals both at the start and end of the assessment period are shown in Table 3.

TABLE 3 Average Health Values :t SD for the e 3 calves Core Temp C Clinical Score WBC X N/L ratios 1000/pl March 13 ave 102.8 3.18 12 0.273 so 0.65 1.15 12 0.19 April 1 ave 102.6 2.95 9 0.129 SD 0.98 1.75 2 0.101 Infrared Values In excess of 20,000 thermal data points were collected on the 65 calves over the two week ment period using the automated thermal station located at the cattle water system. The average radiated temperature value for all calves during this period was lly between 33—35°C. The l radiated thermal value for the true-negative caives for the entire observation period was 34.7“(3 :t 057°C and the value for true—positive calves 35.4OC C (see Figure 9).

Live behavioural observations were also med in this Example 3 to increase the understanding of why true—positive calves generated greater numbers of IRT images (when all images were included) compared with true- negative calves. These obsen/ations suggest that shifts in posture or stance, eye , leg nts, tongue , and ear twitches may cause the lRT pixels to recalculate contrasts, thereby determining it is time to take a new image.

With respect to the data per se, the calves used in the present study were comparatively low stress and expressed a low incidence of BRD of approximately 17%. Of interest however, was the observation that conventional industry standard practice of using pen checking as the primary tool for fying BRD would have identified only two animals and even with the addition of core temperature data at the time of processing only four animals were identified as at risk of BRD and of those, two were subsequently identified as being false positive identifications. in other words, once again, one of the primary challenges with conventional pen checking or clinical score methods for detecting BRD is with the incidence of false negatives.

Example 4 ol Data Salivary and serum cortisol analysis were performed for all animals in the foregoing Examples 1 — 3 (see Table 4). An ELISA assay system, developed at Lacombe Research Centre, was utilized.

In all three Example data sets the cattle were identified as true positive (TP) for bovine respiratory disease (BRD) or true negative (TN) using the hematology, core temperature and clinical score criteria fied in each of the foregoing method sections. Least squares analysis (t—tests) for the cortisol data has also been performed.

Cortisol assays were conducted on blood and saliva samples collected when the cattle arrived at the Lacombe Beef Research Centre and when an animal was identified as t for BRD. Cortisol data displayed erable ion both within and between the studies performed in Examples 1 — 3. Some of this variation is likely due to variation in animal populations, procurement ures and animal history among the groups. There is also likely to be some variation in stress tibility across cattle groups from experiment to experiment.

Table 4 represents overall averages for the animals, without correction for the present data set for animals displaying health aberrations for non—BRD reasons such as transport stress, mechanical insults such as ss or other metabolic reasons such as ation. There were a few of these animals identified and they could arguably cause some bias in the data set. eless, cattle identified as TP for BRD also show a trend towards or an actual statistical increase in cortisol values. Some animals observed also tended to fall into an intermediate group for BRD identification. Again, as described in the methods section, a true negative animal would have displayed a score value of O or 1 for temperature values > 40°C, WBC counts of > 10 or <7 X 103/pL, an N/L ratio of < 0.1 or > 0.8 and a clinical score of < 3. A true positive animal would display a value of 3 or 4 of these criteria and the intermediate animals would display a value of 2. As with the other laboratory criteria, these intermediate s also tended to display an intermediate cortisol value (data not shown).

TABLE 4 Salivary and serum cortisol values in weaned and receiver calves identified as true positive (TP) or true ve (TN) for BRD Data Salivary P Value Serum P value Year Cortisol Salivary Cortisol Serum nmol/L Cortisol nmoi/L Cortisol ' 1 Mean Std. Mean Std. Dev l Dev. 2007 TP 3.16 3.0 NSD 70.96 19.7 P: 0.1 1T—test 2tail, 0.05 2007 TN ‘ 2.95 3.2 52.4 34.9 2008 TP 5.82 2.86 P=0.01 139.7 87.8 P=0.1 1T, 0.5 2T 2008 TN 3.92 1.27 113.3 31.1 2009 TP 2.97 2.47 P=0.06 123.6 61.2 P=0.1 1T, 05. 2T 2009 TN 2.05 1.26 103.3 36.1 These results demonstrate that animals displaying BRD demonstrated a higher infrared radiated temperature and a higher degree of variation associated with that temperature. The cortisol data is also consistent with this g g that BRD animals lly display a higher cortisol vaiue with greater vanafion.

Example 5 Fidget Value and growth efficiency As an alternative to the methods discussed in the Background section above, ce has been ed that demonstrates the use of infrared thermography to classify animals into more efficient and less efficient growth categories (Schaefer, A. L., Basarab, J., Scott, S., Colyn, J., McCartney, D., McKinnon, J., Okine, E. and Tong, A. K. W. 2005. The relationship between infrared thermography and residuai feed intake in cows. J Anim Sci 83(Suppl. 1):263). it is known that animais that are more efficient at growth can display a lower heat loss to the environment. However, the components that make up or account for this difference in efficiency and energy loss are less apparent. To this end, the present apparatus and methods can be used to show that animal behaviour or so called “fidgeting” can be partially responsible for this differential energy use. As such, measuring these “fidgets” would have utility in differentiating animals with different growth efficiency. This Example 5 is provided as a non-limiting example of enting this ple.

Eight crossbred mature cows were used in the present example to test whether a fidget ement also ranked with both a measure of growth efficiency (Residual Feed intake, RFI) and a measure of energy loss (infrared graphy). The cattle were fed a balanced alfalfa cube based diet which met 1.25 times the maintenance nutritional requirement for these animals. The cows were housed in outdoor pens with free access to fresh water and a straw bedded area.

The relative growth efficiency for these animals, referred to as the residual feed intake, had been previousiy determined using a feed bunk monitoring system to record exact feed consumption and weight gain as described by b et al (Basarab, J. A., McCartney, D., Okine, E. K. and Baron, V. S. 2007. Relationships between progeny al feed intake and dam productivity traits. Canadian Journal of Animai Science 87(4):489—502: Basarab, J. A., Price, M. A., Aalhus, J. L., Okine, E. K., Sneiling, W. M. and Lyle, K. L. 2003. al feed intake and body composition in young growing cattle. Canadian Journal of Animal Science 83(2):189—204).

For infrared and fidget measurements the cows were monitored postprandial between a 24h feed . In other words the animals were off feed during the time of monitoring. However, all animals had free access to a water station and when the cows attended the water n they triggered an infrared scanning system which inciuded a radio frequency identification tag (RFlD) thereby recording both their station attendance frequency and their facial ed 3O characteristics.

The daily average values for the infrared scans for the half of the animals with the lowest efficiency and the half of the animals with the highest value are summarized and shown in Table 5 along with the known RFI feed efficiency values for the cow group of 8 animals. The RFI values basically report what an dual animal’s actual feed intake was (as measured by the bunk monitoring systems) compared to what would be a predicted feed intake for that animal based on her body weight and growth. For example, as explained by Basarab et al ( 2003, 2007) an animal with an RFI value of 4 represents a cow that ed 1 kg per day less feed than what would be expected and an RFI value of +1 ents an animal that consumed 1 kg of feed more than what would be predicted. Lower RF! values can represent more efficient cows. in addition as shown in Table 5, the fidget values or number of thermal station triggering events were also collected. This data is expressed as the total number of fidgets per animal per day within four minute drinking bouts.

Using conventional ranking statistics (Spearman Ranking: Tuckman. 1978. Conducting Educational Research, Second Ed. Harcourt Brace Jovanovich Inc. New York) the s in the present study displayed a significant (P<0.05) rank order of thermal values against known RFl values. The animals with the lowest thermal values also yed the lowest RFI values and the lowest number of fidget events. This e demonstrates that a fidget value can have y in identifying animals displaying different production (growth) efficiency.

TABLE 5 Comparative values for RH and IRT in two groups of mature cows Category Mean RFI Mean lRT Mean Fidget Spearman Value Value Value rank value of 0 C Total/day/4 fidget with RFl min drinking bout Higher 4 .3 10.6 4 F5<O.O5 efficiency (low RFI) four cows Lower 0.41 13.2 9 ency (high RFl) four Cows Example 6 Fidget value and estrus it is known in the art that there is a link between restless behaviour and estrus in animals.

Behavioural estrus indicators are the primary means in which producers determine whether dairy cows are in estrus (ie have ovulated or are ready to ovulate). Behavioural indicators of estrus include increased activity such as ng events, pacing/walking, as well as l restless behaviour (eg. lying 1O down and standing up, walking, stepping, and shifting, but also includes many other more subtle behaviours; Poilock, W. E. and Hurnick, L. F. 1979. Effect of two confinement systems on estrus 436 detection and diestrus behaviour in dairy cows. Can. J. Anim. Sci. 59: 799—803.; Walton, J. S. and King, G. J. 1986.

Indicators of estrus in Holstein cows housed in 468 tie stalls. J. Dairy Sci. 69: 973).

Cows housed in tree—stalls exhibit 4 times more activity and restless behaviour during estrus (Kiddy, C. A. 1977. ion in al activity as an indication of estrus in dairy 414 cows. J. Dairy Sci. 60: 235-243). Cows housed in tie~stalls exhibit 2.75 times more activity and restiess behaviour during estrus (when compared with cows not in behavioural estrus). Similar findings have been reported when pedometers were used to e activity and ss behaviour during estrus in free-stalls (Roelofs, J. B., van Eerdenburg, F. J. C. M, Soede, N.

M. and Kemp, B. 2005. 451 Pedometer readings for estrous detection and as predictor for time of ovulation in dairy 452 cattle. Theriogenoiogy. 64: 16904703; Roelofs, J., LOpez-Gatius, F., Hunter, R. H. F., van Eerdenburg, F. J. C. M. and 447 Hanzen, C. 2010. When is a cow in estrus? Clinical and practical s. 448 Theriogenology. 74: 327-344). While pedometers on cows continuously tie stalled have been unable to detect behavioural estrus based upon walking activity measures alone (Feiton, CA, Colazo, M.G., Ponce-Barajas, P., Bench, C.J., and Ambrose, D.J. 2012. Dairy cows continuously—housed in tie—stalls failed to manifest activity changes during . Can. J. Anim. Sci. (in press», the use of a more subtle behavioural biometric of restless behaviour has the ability to capture behavioural estrus even in confined cows. e the fidget biometric, as described within the t specification, can be an accurate and reliable measure of restless our when an animal is standing in a confined space, the use of this type of fldget measure also has the means of capturing restless behaviour exhibited during estrus. As such, the apparatus and methods described herein can identify reproductive states such as estrus.

Example 7 Further fidget data (EDOB/OQ sample) included are calculations on a calf bovine respiratory disease (BRD) data set referred to as ED08/09 (calves analysed in 2008 and 2009).

Briefly, a true positive (TP) animal was one that displayed 3 or 4 out of 4 for a high white blood cell count, a high neutrophile/Iymphocyte ratio, an elevated clinical score and an ed core (rectal) temperature. These criteria are defined in ations known in the art. By contrast, a true negative (TN) animal was one that yed a score of either 0 or 1 out of 4.

Of the ED08/09 data set for 21 animals, 11 of the calves met a TP criteria and 10 met a TN criteria. in other words the prevalence of BBB in this data set was 52%. This result is very similar to the multiple sourced, commingled orted and weaned calves studied eleswhere and is typical for calves of this type in general.

The biometric data collected from the ED08/09 animals for predicting earty disease onset included the absolute infrared value for the eye maximum determination, the mean ratio of the individual calf eye m value compared to the group mean maximum value, the so called MR value, and thirdly, the fidget value for those animals calculated from the same infrared image data set. The five minute s/bout/calf/day information was used. The data used was for the day the animals were verified as TP or TN (so—cailed pull day) and the four days prior to that time.

One approach to determine the relative contribution a given data set has to an overall prediction or ranking of variables is to use a le regression approach and also a discriminant analysis imes called stepwise sion) or logistic regression analysis. Different statistical programs will use different names, for example SAS uses discriminant analysis and MedCalcTM uses the term logistic regression.

Using the ED081059 data, the value for correct TP vs, TN identification using a single biometric measurement was between 57—68%. However, ing all three biometric measurements raised the overall correct identification of animals into disease class ( both TP and TN) to 83.5% . This improvement in correct classification is significant and also offers the ability to identify BRD before the pull day unlike prior art methods.

One method for ranking the relative importance of each biometric measurement in a multi-regression model is to obtain the r value (correlation value), square this value and ly by 100 to obtain the ve percentage importance of a biometric measure or the proportion of the variance that a particular ric value can account for. For example, with a situation like calving difficulty in s (dystocia) it has been determined that the relative ranking of the importance of factors would be as foilows; birth weight and pelvic width of the heifer = 30%, heifer age 10%, calf birth weight = 7% and so on. With the BRD model above the highest ranking (value for predicting BRD onset) for all days was for the orbital absolute infrared vaiue at 33%, next was the fidget value at 16% and the MR value at 10%.

The scope of the Claims should not be limited by the preferred embodiments set forth in the examples, but should be given the broadest retation consistent with the description as a whole.

Claims (17)

What we claim is:

1. An apparatus for identifying biological states in an animal, the tus comprising: means for identifying the animal to obtain animal identification information; 5 at least one infrared thermography camera for photographing the animal to obtain infrared thermography information and fidget information about the animal; and a processor for receiving and processing the animal identification information, the infrared graphy information and the fidget information to determine a biological state of the animal, wherein the biological state is a predictor of onset of 10 disease, growth states and reproductive states in the animal.

2. The apparatus of claim 1, wherein the biological state is selected from the group consisting of a e state, a non-steady state growth period, the onset of puberty and the onset of estrus.

3. The apparatus of claim 1 or claim 2, r comprising an ure accessed 15 by the animal.

4. The apparatus of any one of claims 1 to 3, wherein the infrared graphy information and the fidget ation are obtained postprandial.

5. The apparatus of claim 4, wherein the postprandial period is between a 24 hour feed period. 20

6. The apparatus of any one of claims 1 to 5, wherein the means for identifying the animal comprises antennae for receiving radio-frequency identification (RFID) ation from the animal.

7. The apparatus of any one of claims 1 to 6, wherein the camera is capable of obtaining at least 1 – 60 images per . 25

8. The apparatus of any one of claims 1 to 7, wherein the processor receives the animal identification information, the infrared thermography information and the fidget information wirelessly.

9. The apparatus of any one of claims 1 to 8, wherein the apparatus is automated.

10. A method of identifying biological states in an animal, the method comprising: identifying the animal; photographing the animal to obtain infrared graphy images and fidget information from the animal; and 5 processing the infrared thermography images and the fidget information to identify a biological state of the animal identified, wherein the biological state is a predictor of onset of disease, growth states and reproductive states in the animal.

11. The method of claim 10, wherein the method can be performed during a postprandial period. 10

12. The method of claim 11, wherein the postprandial period is between a 24h feed

13. The method of claim 10, n the method is automated.

14. The method of any one of claims 10 to 13, wherein the images are obtained from or around the l area of the animal.

15 15. The method of any one of claims 10 to 14, wherein the biological states is selected from the group consisting of a disease state, a non-steady state growth period, the onset of puberty and the onset of estrus.

16. An apparatus as defined in claim 1 and as ntially described herein with reference to one or more of the accompanying figures. 20

17. A method as d in claim 10 and as substantially described herein with reference to one or more of the accompanying examples.