Depression is a leading cause of disease worldwide. The association between gut microbiota and depression has barely been investigated in the Japanese population. We analyzed Iwaki health check-up data collected from 2017 to 2019 and constructed generalized linear mixed models. The independent variable was the relative abundance of each of the 37 gut microbiota genera that were reported to be associated with depression. The dependent variable was the presence of depression assessed by the Center for Epidemiologic Studies Depression Scale. Potential confounders, including grip strength, gender, height, weight, smoking, and drinking habits, were adjusted in the regression models. Nine genera’s regression coefficients (Alistipes, Blautia, Coprococcus, Dorea, Faecalibacterium, Holdemania, Lactobacillus, Mitsuokella, and Oscillibacter) showed statistical significance after multiple comparisons adjustment. Among these nine gut bacteria genera, Alistipes, Blautia, Coprococcus, Dorea, Faecalibacterium, and Oscillibacter were reported to be associated with butyrate production in the intestine. Our results indicate that gut microbiotas may influence the depression condition of the host via the butyrate-producing process.

Digital phenotyping requires continuous, real-world collection of multimodal behavioral and physiological data. Although several smartphone-based sensing frameworks have been developed and used, open-source platforms for long-term smartwatch sensor data acquisition remain limited. To bridge this gap, SensorW was developed, an open-source Wear operating system (OS)-based framework for direct lifelog data collection from smartwatches without requiring a paired smartphone or manufacturer cloud service. SensorW includes a front-end Wear OS application for sensor data collection and a back-end server for application initiation, secure data submission, and storage. The front-end application uses battery-management strategies, such as hybrid Global Positioning System sampling and wake-lock management, to continuously collect multimodal sensor outputs, including location data, physical activity–related metrics, barometric pressure, and device wear status, and submit data to the back-end server. A preliminary technical evaluation of SensorW was conducted with seven participants who wore SensorW-enabled devices continuously for at least seven consecutive days. The technical evaluation demonstrated multimodal data collection during 17–42 days of monitoring, with operational data availability ranging from 0.73 to 1.00 across participants. These findings support the technical feasibility of SensorW as a practical and extensible platform for continuous digital phenotyping using consumer-grade smartwatches.