Using centers of pressure tracks of sows walking on a large force platform in farm conditions for locomotion classification

- Sow locomotion classification was possible by means of a force platform.
- The extracted variables permitted a reasonable association to locomotion quality.
- It would be of interest when taking decisions such as sow culling.

This study examines the feasibility of using a 3.0-m-long, 1.5-m-wide force platform to group into clusters the centers of the ground pressure tracks of sows walking on it. The clusters were created according to variables related to the symmetry and cadence of the sows' locomotion, and permitted an evaluation of its soundness in each cluster. Observations were made in a swine-breeding farm that followed standard swine production practices. In the farm, the sows were moved when farrowing from the gestation stalls to the farrowing crates, and were then returned to the service stalls. On these occasions, as recorded over the course of six months, each sow separately passed through a corridor connecting the two rooms, which is where the force platform was placed. The sows were not trained for this task. Signals were separately extracted from four load cells located under the platform, and were processed to obtain the center of pressure (CoP) and the vertical ground reaction force (F) of each sow as it walked on the platform (322 CoP tracks). The trajectory of each sow was derived from the generated CoP track. A gait cycle was considered complete when the CoP track oscillated (swayed) once in the plane of transversal of the sow's trajectory. In each gait cycle, the following variables were calculated: mean velocity, normalized impulse balance per gait cycle, number of relevant peaks of F per gait cycle, and peak ratio obtained from the autocorrelation function of F. Using these variables, all CoP tracks were classified into three clusters (p < 0.05). The relationships among the variables in each cluster allowed for distinction among the CoP tracks in terms of the soundness of locomotion. No significant differences in the measured variables were observed between the CoP tracks of primiparous and multiparous sows, whereas sows entering the farrowing crates were found to walk more slowly (p < 0.05) and with less balance (p = 0.063) than when leaving it. Considering intraclass correlation coefficient of the variables per CoP track as an indicator of locomotive soundness, the cluster of the fastest and most balanced CoP tracks yielded significantly more reliable impulse balance (p < 0.05) responses than did the other clusters. More reliable impulse balance was also observed in CoP tracks made by multiparous sows than by primiparous ones, and by the sows leaving than those entering the farrowing crates (p < 0.05).