Activities of daily living are closely linked to overall health. The advancement of wearable technologies enables individuals to monitor their health status by tracking various activities, such as heart rate and movement, through devices like smartphones, smartwatches, and smart glasses.

By analyzing data collected from these wearables, machine learning (ML) models can provide early warnings for cognitive and chronic diseases, mental illnesses, and other co-morbidities, enhancing preventative healthcare.

Health monitoring data typically raises privacy and security concerns. Additionally, when collected from isolated homes on a small scale, it hampers extensive model training and limits performance. Incorporating federated learning into in-home health monitoring systems facilitates the joint training of a shared global ML model.

This is achieved under the coordination of a central server that aggregates locally-computed updates while keeping all sensitive data on local clients, such as smartphones. This not only enhances the model's performance but also captures fine-grained information about individual users for personalized healthcare.