Tencent Health AI Clinical Assistant: Clinical Note Writer

Problem

Clinical documentation often starts from fragmented inputs: dictated key points, pasted notes, or files that must become standardized outpatient records, admission records, progress notes, and discharge summaries. This Skill targets clinicians in outpatient and inpatient settings, converting dictation, pasted text, or .txt, .pdf, and .docx files into a compliant medical record draft.

How It Works

• Draft first, then complete: the first response provides a draft; missing fields are marked [missing], inferred values are marked [inferred], and pending ICD codes are marked [pending ICD].
• Terminology and follow-up: it normalizes colloquial terms such as “stomach ache” to “abdominal pain”, detects missing items, and returns no more than three actions per turn; clinicians can answer, skip, or finalize.
• Safety first: conflicts such as allergy versus medication contraindication are flagged with 🔴 and resolved before further completion.
• Template adaptation: hospital or department templates can be uploaded, then content is reorganized by section order, field names, headers, and signature blocks while preserving values.

Boundaries

It does not perform diagnostic reasoning, differential diagnosis, treatment planning, or medication adjustment. Diagnostic conclusions from a diagnosis-support workflow can be mapped into diagnosis sections. It does not create clinical decisions or fabricate data, and patient information is not cached or transmitted externally. Use it for reviewable record drafting, not as final medical judgment.