Parkinson's & Epileptic Behavior Recognition

Problem

At-home episodes of Parkinson's disease or epilepsy often depend on manual supervision and retrospective recollection, making it hard to continuously capture tremor, convulsion, stiffness, and gait changes. Surveillance video can cover daily activity, but raw footage does not directly produce reviewable observation clues, nor does it easily track a patient's historical state.

How It Works

The skill accepts video or image input, invokes the built-in analysis script, and returns a structured report with risk notes and a report link. Core capabilities include tremor recognition, convulsion recognition, muscle stiffness detection, gait abnormality recognition, and episode statistics.

Typical workflow:

• Provide a local file path or public URL using --input / --url.
• Run scripts.parkinson_epilepsy_behavior_recognition_analysis from the skill root.
• Use --detail to control output depth and --output when saving results is needed.
• Query historical reports with --list; results should be read from the cloud API and rendered as a table.

Boundaries

This skill is intended for at-home supportive monitoring and does not replace professional medical diagnosis. Cameras should be fixed and well lit, with the patient's full body or upper body visible in the frame. Supported formats are jpg/jpeg/png/mp4/avi/mov, with a maximum file size of 10MB. If frequent abnormalities are detected, seek timely medical review and adjust the treatment plan.