Author:
Wassermann Laura,Helgers Simeon O. A.,Riedesel Ann-Kristin,Talbot Steven R.,Bleich André,Schwabe Kerstin,Häger Christine
Abstract
In animal experimentation, welfare and severity assessments of all procedures applied to animals are necessary to meet legal and ethical requirements, as well as public interests. So far, the methods suggested for this purpose are time consuming and personnel intensive. Also, evidence-based biostatistical methods for this purpose are still rare. We here tested whether the classification of heart rate (HR) and activity (Act) data monitored by telemetry in the home cage by unsupervised k-means-based class-labeling and subsequent Support Vector Machine (SVM) analysis allows severity assessment and grading of experimental procedures of different domains, including surgery, injection, behavioral testing, and routine handling for maintenance. Telemetric devices were subcutaneously implanted in young adult male Crl:CD(SD) and BDIX/UImHanZtm rats. After recovery, rats were randomly subjected to different experimental procedures, i.e., handling and cage change as routine maintenance, Rat Grimace Scale, burrowing, and social interaction for welfare assessment, as well as repeated subcutaneous injections. Thereafter, rats were either intracranially implanted with electrodes or injected with tumor cells. Directly after each procedure, HR and Act were monitored by telemetry in the home cage for 4 h. Application of k-means and SVM algorithms on the obtained data sets from baseline (as no stress), cage change (exploratory Act), and intracranial surgery (as burden) measurements computed three classes described as low HR/low Act, high HR/high Act, and high HR/low Act, respectively. Validation of the SVM model by entering data from all procedures confirmed the allocation to the high HR/low Act class (burden) after surgery, which lasted longer after subcutaneous transmitter implantation than after intracranial surgery. The majority of data points from repeated injections, behavioral testing, and maintenance handling were allocated to the low HR/low Act and high HR/high Act classes. Overall, the SVM model based on HR and Act data monitored in home cage after procedures may be useful for the classification and grading of experimental procedures of different domains.
Funder
Deutsche Forschungsgemeinschaft
Cited by
8 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献