Top Navigation Bar

ASAE Conference Proceeding

This is not a peer-reviewed article.

Early Detection of Calf Diseases by Automatic Recording of Behavioral Changes

T. Eberhardt, H. Grimm, T. Richter, P. Savary

Pp. 350-355 in Fifth International Dairy Housing Proceedings of the 29-31 January 2003 Conference, (Fort Worth, Texas, USA), ed. K. A. Janni. ,Pub. date 29 January 2003 . ASAE Pub #701P0203

Abstract

Total calf losses during rearing can be up to 15% during the first 6 weeks of life. Calf diseases cause high costs for veterinary treatment, lower growth rates of the calves, and are a problem for animal welfare. Continuous animal supervision is of great importance, but almost impossible with larger groups. Modern automatic feeders facilitate feeding of young calves, allow a more physiological uptake of milk (replacer), and give us the chance to record their behavior during feed uptake.

Aims of the study were to check if established (uptake of milk, number of visits at the feeder station) or new parameters give indications that the animal is undergoing some infection. These new parameters were behavior during sucking - drinking time or velocity, pauses during sucking, butts with head against the teat - or after sucking - sham sucking, changing the sucked teat, butts. Vacuum level at the artificial teat(s) and movements of these teats were recorded (50 measurements/sec.) online and used to calculate traits. A veterinarian and the responsible person checked health status twice per day. Behavior after drinking, i.e. when the supply of liquid milk replacer was shut off automatically after intake of 2 to 3 L milk, showed significant differences between "healthy" and "ill" days of the same calf. Even one or two "day(s) before ill" most calves showed changes in behavior (although not always significant). Indication of health status by these parameters may depend on illness type, especially if the calf suffers from fever before or at the observable beginning of the illness. Observations and development of calculation programs will be continued. If calves can be detected "ill" at such an early disease state, treatment costs may be reduced and calf welfare enhanced.

KEYWORDS. Calf, Calves, Behavior, Automatic, Detection, Diseases.

Introduction

Calf diseases are one of the most important problems in dairy husbandry, causing lethal losses of about 10% (15%) during week 1 (1-6) of life (Platen, 1999). The earlier any disease is detected, the earlier veterinary treatment can commence and the better are the prospects for complete recovery of the animal. Many diseases start with behavioral changes before any other symptoms are detectable. It is difficult for the farmer to see these changes in a bigger group of calves and it is even more difficult when feeding of the calves is automated. So we need automatic feeding systems capable of automatically detecting these behavioral changes during the visits of the calves. Examples for changed behavior are less uptake of milk or milk replacer (Rademacher, 2000), slower uptake (Maatje et al., 1993), as well as lower vacuum during sucking and less change of teats, when two artificial teats are offered (Meisinger, 1998).

So the objectives of this investigation are to see if behavioral traits change before other symptoms can be observed by the herdsman.

Animals, Material, Methods

Between March, 2000 and February, 2001 sixty calves - male and female, all from the producing dairy herd of the research station - were observed and their feeding behavior was automatically recorded whenever they came to the feeding station. Feeding (max. 8 kg milk replacer /day with a concentration of 100 g milk powder/kg water) was given by an automatic feeder (Stand-Alone; manufactured by Förster-Technik, Engen) coupled with an animal scales. Concentrates were supplied by an automatic feed dispenser (Förster-Technik).

Traits of sucking behavior were recorded continuously using pressure transducers (Keller Druckmesstechnik; Winterthur, Switzerland) and butts to the teats during or after sucking were recorded with a distance potentiometer. Figure 1 gives an impression of the system. Automatic data acquisition was done with a plug-in PCI 20000 Series Multifunction Board together with Visual Designer ™ Software, Version 3.0 (Intelligent Instrumentation, Inc., Tucson, Arizona). All behavioral data during sucking were recorded at a rate of 50 cycles/s and stored in regular intervals on a cd-rom for documentation. After data acquisition, intensive programs were applied to reduce the amount of data and to obtain traits for statistical calculations. These calculations resulted in one set of data for each visit at the feeder station. The traits were: sucking duration, and vacuum (min/max); sucking frequency; length and number of pauses during sucking; intensity and number of butts; number of changing teats; and length of visit.

Figure 1: Additional parts of the feeding station

350-355_files/image1.gif

Signs of illness were recorded by a veterinarian (first author). She looked after the calves 1 to 2 times each day and measured rectal temperature also. The herdsman measured temperature and looked at the calves `as usual´. According to the symptoms, five major groups of illness were reported:

  • Fever

  • Pneumonia

  • Diarrhea

  • Otitis

  • Divergent abdomen (?div. abdomen”)

  • All diseases (“All”)

Most cases of “otitis” (shaking the head, hanging ears, strange sound inside when the ears were massaged) were found some days after the end of diarrhea, starting with changes of behavior and higher temperature (39 – 39.5 °C) already one to two days before final diagnosis. “Divergent abdomen” was the diagnosis of an unusual abdominal sound, full abdomen, acid rumen contents, or high tension of abdominal skin. This group also includes colic.

Most of the interest was to test the hypothesis that changes in behavioral traits occur before an illness is detected, so comparisons were made according to the description of days before and during illness in table 1. These comparisons were:

healthy/ill: h4 & h3 versus i1 & i2 (see table 1)

2before/ill: h2 versus i1

1before/ill: h1 versus i1

2before/1before: h2 versus h1

Table 1. Description of days for the comparison of behavior before and during illness

healthy

healthy

ill

days before first symptoms are observed

days when symptoms are observed

h6

h5

h4

h3

h2

h1

i1

i2

i3

i4

All traits were calculated from the vacuum curves during all visits at the feeder station, either “during” sucking on a teat with milk flow or when no milk was available. The latter was preset if the interval since the last visit was too short - i.e. less than 2-4 h, depending on the amount of available milk per day. When no milk was available calves normally left the station after a short time, after intensively checking one or two teats that no milk was available. Traits of this behavior are marked “after”. Here also traits of behavior after end of milk supply - when the electronic valve was shut off - are included. In this situation healthy calves are sucking and butting stronger than during sucking with milk supply.

Statistical calculations were made to detect differences in behavior before and after observation of signs of illness. These were done with non parametrical tests after calculating one mean value per calf and day. If these comparisons showed significant differences, all sets of data with the respective diagnosis were analyzed further (SPSS 10.0 Software with mainly the module `General Linear Model – Univariate’).

Results

Some of the calculated traits were not different between “ill” and “healthy”, other traits proved to be suitable for differentiation. In table 2 the different traits between ill and healthy are listed if the same pattern of changes was observed for more than one diagnosis. For example, the intensity of butts “after” trait was significantly lower as indicated by the ? when the animals suffered from fever, div. abdomen, or otitis two days before illness was detected (2before/ill). However, the intensity of butts “after” trait was significantly ? when the animals suffered from pneumonia two days before illness was detected (2before/ill). The signs ? or ? indicate direction of change and significance.

A clear tendency was observed that the animals either suck weaker or pause to breath more often (pneumonia) when they are suffering from the observed diseases. Changing teats “during” was significantly higher on healthy days. Although this was very consistent for almost all observed cases, this trait was not suitable for early detection, as it occurred only after the animals were already recognized as ill. More differences in other traits were observed, being significant for a single diagnosis, however.

Figure 2 gives an example of the recorded sucking and butting "behavior" of one calf at the feeding station. Most ill calves pause more often or longer during sucking and butt less often or weaker when milk flow ceases.

Conclusion

Early detection of behavioral changes is possible before other signs of illness are seen. It is difficult or almost impossible, however, to draw a clear conclusion that individual animals are really ill. This cannot be achieved by automatic evaluation of these data alone, however, they can be helpful as a `warning´. Veterinarian experience cannot be replaced. In this study, a veterinarian continuously looking after the animals with more skill than a “normal” person diagnosed first signs of illness. If this “normal” person could detect the diseased calf only one day later the differences and certainty of automatic early detection would be even better. Assuming only this single day delay for recognition of the disease by the herdsman gives a good idea of the advantage of automatic recognition of diseases. This gives a good chance for earlier and more intensive care as well as veterinary diagnosis and treatment.

Most traits can be calculated from the signals of a relatively cheap pressure sensor. The calculations will be possible without sophisticated software, as the principal connections are known now. The detection of number and intensity of butts makes a separate sensor necessary. More intensive investigations, however, are needed to determine if this measurement is required.

REFERENCES

Maatje, K.; J. Verhoeff; W.D.J. Kremer; A.L.M. Cruijsin; T. S. G: A. M. van den Ingh 1993. Automated feeding of milk replacer and health control of group-housed veal calves. Vet. Rec., 133, 266-270.

Meisinger, I. 1998: Veränderungen im Saugverhalten von Kälbern am Tränkeautomaten bei Erkrankung. Diploma thesis, University Hohenheim

Platen, M.; F. Werner; A. Reiter; H.-J. Löhnert 1999 Kälberverluste: Von 18 Prozent auf nahe Null. Top Agrar 9, R20-22

Rademacher, G. (2000) Kälberkrankheiten. Verlagsunion Agrar

Table 2. Significance and direction of behavioral changes during onset of illness

Trait

healthy/ill

2before/

ill

1before/

ill

2before/

1before

Diagnosis

?

“all“

?

fever

intensity of butts ”after“

?

pneumonia

?

div. abdomen

?

otitis

?

“all“

?

fever

duration of sucking after supply of milk

?

pneumonia

?

?

diarrhea

?

div. abdomen

?

?

“all“

?

fever

N of butts “after“

?

?

?

pneumonia

?

diarrhea

?

“all“

N of changing the teats “during”

?

fever

?

diarrhea

?

div. abdomen

?

?

fever

Duration of pauses “during”

?

pneumonia

?

div. abdomen

?

?

otitis

?

“all“

Duration of visit

?

?

pneumonia

?

?

diarrhea

?

“all“

Waiting after milk is offered

?

fever

?

?

?

diarrhea

The signs ? or ? indicate direction of change and significance

Figure 2. Sucking and butting of calf 14 - most ill calves pause more often/longer during sucking, and are butting less/weaker when milk flow ceases (top: healthy; bottom: otitis). Y-axis is calibrated in kPa, different shades of gray after end of milk supply indicates the measurements at the two teats (n of changes!); X-axis divided by 50 seconds.

350-355_files/image2.jpg 350-355_files/image3.gif