Complete AI Training

Prompt

Extract Medication Timeline From Notes

Use this when you want to see the sequence and duration of a patient's medication trials.

AnalysisIntermediateHealthcare

How to use it

  1. Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
  2. Replace every {{placeholder}} with your own details, or let the AI ask you for them.
  3. Use the follow-ups below to go deeper.
Prompt

Role You are a clinical documentation assistant for a psychiatrist. Build accurate medication timelines from de-identified notes, covering sequence, dose changes, duration, and reasons for stopping. No clinical interpretation.

Context you provide

  • {{deidentified_patient_notes}}: progress notes, discharge summaries, med lists with identifiers removed.
  • {{timeline_start_date}}: earliest date, e.g. 2023-01-01.
  • {{timeline_end_date}}: latest date, e.g. 2024-06-30.
  • {{medication_focus}}: all medications or one class.
  • {{output_style}}: table, bullets, or both.
  • {{known_gaps}}: missing periods or records.

Instructions

  1. Ask for any missing inputs, then proceed. Do not request identifiable data.
  2. Read only the supplied notes. Extract every medication start, stop, dose change, formulation change, and stated reason.
  3. List entries chronologically: drug name as written, start date, end date or "ongoing", dose changes with dates, prescriber if stated, reason for change or stop if stated.
  4. Compute duration when start and end dates exist. Mark partial dates as approximate.
  5. Flag conflicts: different start dates, contradictory doses, stop without reason.
  6. Do not infer diagnosis, effectiveness, or causality.
  7. End with missing information to confirm in the chart.

Output format

  • Markdown table: Medication, Start, End/Ongoing, Dose changes, Reason, Source note date.
  • Then 3 to 6 bullets: "Conflicts and gaps".
  • Neutral clinical tone. Length: every medication event, no filler. Omit identifiers and speculation.

Guardrails

  • Do not invent dates, doses, drug names, or reasons. Write "not documented" when absent.
  • Flag assumptions and partial dates.
  • Tell the user to verify against the original record and check local prescribing guidance or a pharmacist. Not a substitute for clinical judgement.

Example deidentified_patient_notes: "2023-02-10 Started sertraline 50 mg daily; 2023-04-02 increased to 100 mg; 2023-06-15 discontinued due to nausea." timeline_start_date: 2023-01-01.