Evaluation of Broiler Growth Velocity and Acceleration in Relation to Pulmonary Hypertension Syndrome W. B. Roush*,1 and R. F. Wideman, Jr.† *Departments of Poultry Science, The Pennsylvania State University, University Park, Pennsylvania 16802-3501 and †University of Arkansas, Fayetteville, Arkansas 72701 tively). The hypothesis was accepted that normal birds have greater oscillations in growth velocity and acceleration than birds susceptible to PHS. A general regression neural network (GRNN) with genetic adaptive calibration was trained to predict PHS based on individual growth phases and their combinations. Data representing the first, first two, and all three phases of growth were determined to have potential for computerized diagnostic weighing. With the GRNN, birds in all three data sets were successfully classified (100%) with or without PHS. A third hypothesis, therefore, was accepted that artificial neural networks could be used to distinguish the difference between normal broilers and those susceptible to PHS. In the second experiment, only one bird was diagnosed with PHS. Velocity and acceleration neural networks from Phase 1 and Phases 1 and 2 in the first experiment were applied to the growth velocity and acceleration data of Experiment 2. The Phase 1 neural networks were the most promising in that they correctly identified 71.6 and 72.4% of the birds as normal for velocity and acceleration data, respectively. In general, data in the second experiment exceeded the neural network range of training for both velocity and acceleration, which reflected increased oscillation during the second phase of growth.
ABSTRACT An evaluation was made of the relationship between individual daily growth patterns and susceptibility of broiler chickens to pulmonary hypertension syndrome (PHS). In the first experiment, 46 male broilers were weighed for each of 50 d, during which time 13 developed PHS. Three temporal phases (0 to 15, 16 to 35, and 36 to 50 d) of broiler growth velocity and acceleration were examined. Correlation dimensions and Lyapunov exponents suggested evidence of chaos in growth velocity and acceleration, but the absence of detectable differences between broilers in the normal and PHS categories led us to reject the hypotheses that growth is more chaotic in normal broilers than in broilers susceptible to PHS. Growth velocity and acceleration values for mean and SD were statistically evaluated as response variables for each growth phase. Mean values for velocity during the third phase were different between broilers in the normal and PHS categories (velocity: 68.8 vs 48.9 g/d, P = 0.03, respectively) and (acceleration: 0.3 vs –1.4 g/d2, P = 0.07, respectively). The third phase SD (reflecting oscillation for velocity and acceleration) was greater for normal than for PHS birds (velocity: 26.1 vs 21.3 g/d, P = 0.13, respectively; acceleration: 39.7 vs 28.2 g/d2, P = 0.03, respec-
(Key words: growth velocity, growth acceleration, pulmonary hypertension syndrome, artificial neural network) 2000 Poultry Science 79:180–191
1996), suggesting that other characteristics related to growth may be more predictive of susceptibility to PHS (Roush et al., 1994). In addition to absolute cumulative daily body weight gain, growth can be described in the terminology of physics as volumetric motion in time, which can be viewed in terms of velocity and acceleration. May (1976) has shown that as mathematical rates (as coefficients) are increased in a difference equation (e.g., a logistic equation), the mathematical output evolves from static to periodic to aperiodic responses. Roush et al. (1994) demonstrated periodic and aperiodic responses in the
INTRODUCTION Pulmonary Hypertension Syndrome (PHS) is a metabolic disorder that increases in incidence when broiler flocks grow rapidly and decreases in incidence when flock growth rates are restricted (Dale and Villacres, 1988; Reeves et al., 1991; Schlosberg et al., 1991, 1992; Arce et al., 1992; Acar et al., 1995; Tottori et al., 1997). However, it is not necessarily the fastest growing individuals in the flock that develop PHS (Wideman and Kirby, 1995a,b;
Received for publication April 5, 1999. Accepted for publication September 30, 1999. 1 To whom correspondence should be addressed: William B. Roush, Poultry Science Department, The Pennsylvania State University, 220 Henning Building, University Park, PA 16802-3501; e-mail:
[email protected].
Abbreviation Key: PHS = pulmonary hypertension syndrome; GRNN = general regression neural network; PERL = Poultry Environmental Research Laboratory; RV:TV = right ventricular:total ventricular weight; HCT = hematocrit; P1NN = phase 1 neural network; P12NN = phase 1,2 neural network.
180
BROILER GROWTH IN RELATION TO PULMONARY HYPERTENSION SYNDROME
dynamics of day-to-day growth in broilers. This phenomenon showed three temporal growth phases (perhaps bifurcations) (Phase 1, 0 to 15 d; Phase 2, 16 to 35 d; and Phase 3, 36 to 50 d), which are approximately correlated with the sigmoid shape of the cumulative growth curve. Based on the association between the incidence of PHS and flock growth rates, the present study was conducted to evaluate three hypotheses. First, individual normal broilers, as compared with individual birds with PHS, may exhibit aperiodic growth responses as reflected by the mathematics of chaos (i.e., differences in the oscillations of growth velocity or acceleration during one or more of the growth phases.) In cardiac research, Poon and Merrill (1997) have shown heartbeat intervals to be chaotic and have increased variance in healthy individuals as compared with those having chronic heart failure. The second hypothesis was that larger aperiodic oscillations would be detectable by analysis of variance more in normal broilers than in those susceptible to PHS. A third hypothesis was that that differences between normal broilers and broilers susceptible to PHS can be determined based on an artificial neural network analysis of individual daily growth velocity or acceleration over time. Artificial neural networks previously have been used to successfully make diagnoses and predictions of complex biological data (Roush et al., 1996, 1997; Roush and Cravener, 1997, Cravener and Roush, 1999).
METHODS AND PROCEDURES Experiment 1 General. The first experiment was conducted in environmental chambers2 1 and 2 within the Poultry Environmental Research Laboratory (PERL, Building A304, University of Arkansas Poultry Research Farm) using procedures under University of Arkansas IACUC Protocol #241 (Pathology of the Pulmonary Hypertension Syndrome in Broilers). A Peckode3 weigh platform/individual bird I.D. system was installed on the floor of Chamber 1 and was connected by an extension cable to the Peckode Controller/Keyboard and power transformer mounted outside the chamber. The Pecklog software was installed in a Gateway 2000 486DX computer,4 which was reconfigured to accommodate Pecklog software requirements for recording body weight data via Comport 1, bird I.D. via Comport 2, and mouse operations via Comport 3. Stray electromagnetic interference within the PERL initially prevented the detection of the individual I.D. microchips by the Peckode weigh platform. To provide electrical
2 Dimensions: 11′9″ length × 7′6″ width × 7′6″ height = 88 ft2 (8.2 m2) floor space per chamber. 3 NEDAP AGRI BV, PO Box 9, 7255 ZG Hengelo (Gld), The Netherlands. 4 Gateway 2000, North Sioux City, SD 57049-2000. 5 Hubbard Farms, Waldpole, NH 03608. 6 Mettler-Toledo, Inc., Hightown, NJ 08520. 7 Average weight of wing band + transponder = 1.8 g.
181
shielding, metal panels lining the interior of Chamber 1 were connected through heavy gauge copper wire to a copper ground rod driven 3 M into the earth outside the PERL. A new electrical conduit was installed to carry interference-free current directly from the PERL main fuse box to the Peckode system and computer. The Peckode system was calibrated according to specifications and then was evaluated for linearity and reproducibility.
Protocol A floor tray for starter feed and a hanging tube-type feeder for pelleted feed were placed near the front of each chamber, and a Plasson-type waterer was hung at the rear of each chamber. A fence was installed between the feed and water in Chamber 1; the Peckode weigh platform spanned the opening in the fence. Fresh wood shavings litter was provided in both chambers. Chamber temperatures were brought to 33 C (92 F) 2 d prior to the arrival of the chicks. Four hundred male byproduct chicks of the Hubbard 6610 (Hi Yield)5 breeder pullet line were weighed using a Mettler SB 16001 Delta Range威6 electronic balance on the day of hatch (Day 1). Fifty-two clinically healthy chicks weighing 40 g or more were wing banded with Peckode I.D. microchips7 and were placed in Chamber 1. All obvious culls and chicks weighing less than 40 g were euthanized, and the remaining chicks placed in Chamber 2. For the first 9 d, chicks were fed a mash formulated to meet or exceed the minimum NRC (1994) requirements of 23% CP and 3,000 kcal ME/kg; thereafter, the same feed was provided as pellets. The birds consumed feed and water ad libitum throughout the experiment. The waterers were cleaned daily. Chamber temperatures were 33 C (92 F) on Days 1 to 5, 29 C (85 F) on Days 6 to 8, 27 C (80 F) on Days 9 to 13, and 22 to 24 C (72 to 75 F) on Days 14 to 50. The light:dark cycle was 24 h light:0 h dark for Days 1 to 7 and 23 h light:1 h dark thereafter. Throughout the experiment, all birds in Chamber 1 were weighed daily between 0800 and 1000 h using the Mettler SB 16001 electronic balance. Initially, the chicks crossed the Peckode weigh platform too rapidly for weights to be recorded, and many of the chicks did not find their way through the fence opening, possibly because there was no direct line of sight between the feed and water. The chicks were individually introduced to the feed and water and were manually placed on the Peckode weigh platform twice daily on Days 1 and 2. After Day 2, the fence and weigh platform were repositioned to provide a direct line of sight between the feed and water. In addition, a white dome provided by Peckode was attached to the weigh platform to provide the chicks with an imprinting target. Subsequent to these changes, all chicks readily crossed the weigh platform and most paused beside the white dome long enough for their weight to be recorded at least once daily. On Day 20, the gate in the fence was widened, and the white dome was moved laterally to provide more space on the surface of the weigh platform. Also on Day 20, the Peck-
182
ROUSH AND WIDEMAN TABLE 1. Data for normal, pre-PHS, and PHS broilers1 Bird
Category
RV:TV ratio2
O2 Saturation
826 829 833 848 856 863 875 893 900 Mean ± SEM 846 850 868 870 873 895 Mean ± SEM 830 832 836 843 853 871 891 Mean ± SEM
Normal Normal Normal Normal Normal Normal Normal Normal Normal
0.204 0.252 0.214 0.181 0.233 0.181 0.249 0.229 0.197 0.216 ± 0.009 0.308 0.321 0.453 0.335 0.357 0.352 0.354 ± 0.021 0.383 0.282 0.361 0.304 0.444 0.448 0.295 0.360 ± 0.028
80 91 92 85 93 84 91 89 84 88 ± 2 68 74 72 76 67 69 71 ± 1 — — — — — — 75 75 ± 0
(%)
Pre-PHS Pre-PHS Pre-PHS Pre-PHS Pre-PHS Pre-PHS PHS PHS PHS PHS PHS PHS PHS
Final body weight
Day of death
(g) 2,745 2,908 2,191 2,758 2,404 2,673 2,624 2,551 2,460 2,590 ± 72 2,375 2,286 2,176 2,427 2,645 1,876 2,298 ± 103 1,509 1,096 1,427 1,050 2,453 1,911 2,142 1,655 ± 212
50 50 50 50 50 50 50 50 50 50 50 50 50 50 50 42 40 41 35 50 50 50
PHS = pulmonary hypertension syndrome. RV:TV ratio = right ventricular:total ventricular weight.
1 2
ode I.D. microchips were transferred from the wing bands to leg bands to preserve the necessary proximity of the microchips to the weigh platform I.D. detector. At this time, the I.D. microchips from seven birds that died (I.D. numbers 627, 3097, 3690, 3749, 5380, 5579, 8425) and from the smallest birds in the Chamber 1 (I.D. numbers 770, 936, 3293, 4025, 4888, 5552, 5553, 5832, 5887, 9478, 9562) were reassigned to large healthy birds transferred from Chamber 2. The small birds remained in Chamber 1 for manual weighing. Two I.D. chips failed (900 and 8464), and these birds also remained in Chamber 1 for manual weighing. Birds that died after Day 28 were necropsied to determine the cause of death. Pulmonary hypertension syndrome was diagnosed if abdominal fluid accumulation (ascites) was evident, or if a plasma clot adhered to the surface of the liver. For birds surviving to Day 50, the universal C-sensor attached to a Vet/Ox 4403 pulse oximeter8 was positioned on the wing between the radius and ulna to measure heart rate and percentage saturation of hemoglobin with oxygen (Peacock et al. 1990; Julian and Mirsalimi, 1992; Wideman and Kirby, 1995a,b). Blood was obtained by venipuncture for duplicate hematocrit (HCT) determinations using heparinized capillary tubes and a microhematocrit centrifuge. Birds surviving by Day 50 were euthanatized with CO2 gas, and hearts were removed, dissected, and weighed for calculation of the right ventricular:total ventricular weight (RV:TV) ratio as an index of pulmonary hypertension (Burton et al., 1968:
8
Sensor Devices, Inc., Waukesha, WI 53188.
Cueva et al., 1974; Sillau et al., 1980; Huchzermeyer et al., 1988; Peacock et al., 1989). Specific hypertrophy of the right ventricle, as reflected by an elevated RV:TV weight ratio, provides definitive evidence that the right ventricle has performed additional work to maintain an elevated pulmonary arterial pressure leading to PHS (Burton et al., 1968; Cueva et al., 1974; Huchzermeyer and DeRuyck, 1986; Hernandez, 1987; Peacock et al., 1989; Julian, 1993; Odom, 1993; Wideman and Bottje, 1993; Lubritz et al., 1995). Definitive criteria were used to identify the normal, pre-PHS, and PHS groups (shown in Table 1). Birds were considered for inclusion in the groups only if their body weights had been recorded thoughout the 1- to 21-d interval, thereby permitting a consistent evaluation of early growth dynamics. Birds included in the normal (clinically healthy) category had RV:TV ratios of ≤0.252 and oximetry readings of ≥80%. Birds included in the pre-PHS category had RV:TV ratios of >0.280, oximetry readings of
