Artificial intelligence (AI)-based image recognition allows computers to observe experimental mice and guinea pigs in video recordings and automatically identify behaviors such as eating, drinking, climbing, and jumping. AI can also be used to analyze guinea pig vocalizations by converting sound recordings into spectrograms—visual representations that show how sound frequency and intensity change over time. By examining these vocal patterns, researchers can investigate calls associated with different physiological or emotional states, including hunger, agitation, and stress. These technologies provide a faster and more objective way to assess animal behavior and condition, with potential applications in disease research, long-term health mo

This study uses video and audio data to develop quantitative analytical workflows tailored to the behavioral characteristics of different laboratory animal species. Selected patterns of daily activity, environmental interaction, and vocalization are converted into comparable observational measures. Although these data are currently analyzed as separate components, the framework provides a foundation for future multimodal integration. It may complement conventional behavioral tests by capturing changes over longer periods and provide additional evidence for disease-model assessment, evaluation of intervention effects, and animal welfare monitoring in preclinical studies.

This technology may serve as a supporting analytical tool for laboratory animal research and care, providing long-term observational information that is not easily captured by conventional behavioral tests. Standardized processing of video and audio data may reduce part of the manual assessment workload while improving data consistency and traceability. The approach has potential applications in disease-model observation, pre- and post-intervention comparison, and animal welfare monitoring, and may provide a foundation for future preclinical analytical workflows and research services.