# AI medical scribe trial worksheet
Sphygmos | Version 2026-09-16
https://sphygmos.com/guides/ai-medical-scribe-evaluation/

Use fictional records only. This is an original product-evaluation aid, not a clinical validation study, diagnostic benchmark or regulatory certification.
Use the same script and comparable conditions for each supplier. Record failed attempts as well as successful ones. Do not enter patient information in this worksheet.

Supplier / product: ____________________
Product version / date: ____________________
Device / browser / microphone: ____________________
Mode (dictation / conversation recording): ____________________
Language / accent / relevant background noise: ____________________
Participant and recording permission confirmed: ____________________

## Keep the speaker and family history straight
Sample: Fictional patient Lindiwe Test: "My sister has asthma. I have never been diagnosed with asthma." Doctor: "Family history: sister with asthma. The patient reports no previous asthma diagnosis."
Check: The sister, not Lindiwe, has the reported asthma history. The note must not turn a family history into a patient diagnosis.
Observed (met / correction needed / failed / not tested): ____________________
Incorrect or missing facts: ____________________
Record / generation / review-and-correction / save seconds: ____________________
Retry, assistance or another system needed: ____________________

## Keep an unknown medicine unknown
Sample: Lindiwe Test: "I take a tablet for blood pressure. I cannot remember its name or strength." Doctor: "Medication reconciliation is still needed. No new prescription has been issued in this encounter."
Check: The medicine and strength remain unconfirmed. No invented drug, dose, frequency or prescription appears.
Observed (met / correction needed / failed / not tested): ____________________
Incorrect or missing facts: ____________________
Record / generation / review-and-correction / save seconds: ____________________
Retry, assistance or another system needed: ____________________

## Preserve a correction and a negative
Sample: Lindiwe Test: "I said left wrist earlier, but I meant my right wrist. I have not had a fever." Doctor: "Right wrist symptoms; patient reports no fever."
Check: The final note records the right wrist and the reported absence of fever. It does not silently change a reported negative into an examination finding.
Observed (met / correction needed / failed / not tested): ____________________
Incorrect or missing facts: ____________________
Record / generation / review-and-correction / save seconds: ____________________
Retry, assistance or another system needed: ____________________

## Separate a plan from a completed action
Sample: Doctor: "We will request the previous report. It has not arrived. A follow-up is proposed for next week, but no appointment has been booked."
Check: The report is outstanding and follow-up is proposed. Neither receipt of the report nor a confirmed booking is invented.
Observed (met / correction needed / failed / not tested): ____________________
Incorrect or missing facts: ____________________
Record / generation / review-and-correction / save seconds: ____________________
Retry, assistance or another system needed: ____________________

## Open a second patient without carrying the first note
Sample: Leave Lindiwe Test with an unfinished draft. Open a separate fictional record named Peter Test and begin a new encounter. Do not copy any details into it.
Check: Peter has his own identity and encounter. Lindiwe's draft is not inserted into Peter's record, and returning to Lindiwe offers the correct unfinished work.
Observed (met / correction needed / failed / not tested): ____________________
Incorrect or missing facts: ____________________
Record / generation / review-and-correction / save seconds: ____________________
Retry, assistance or another system needed: ____________________

## Find out what survives an interruption
Sample: Use a disposable fictional test encounter. Interrupt the connection or reload during a short sample recording, then reopen the same patient and encounter.
Check: Record what was recovered and what was lost, whether the app explains the state, and whether a retry duplicates notes or usage. Do not assume recovery from a success message alone.
Observed (met / correction needed / failed / not tested): ____________________
Incorrect or missing facts: ____________________
Record / generation / review-and-correction / save seconds: ____________________
Retry, assistance or another system needed: ____________________

## Compare the whole workflow
Complete the same fictional work using your current method. Include reviewing, correcting, saving and locating the final record, not just generation time.
Report number of attempts, median and range. Keep dictation and conversation recording separate. A small sample does not establish clinical safety or accuracy.
Treat wrong-patient content, invented medicines and lost work as separate stop-and-investigate findings; do not average them away in an overall score.

## Questions before a patient pilot
Confirm organisational approval, processing terms, subprocessors, data locations, retention, export, access controls and incident support with the supplier.
Confirm plan allowances, separate dictation/consult recording limits, top-ups and the existing services you still need.
Agree the pilot scope and review process with the responsible clinicians. No patient-data use is authorised by completing this worksheet.

Published by Sphygmos, a software supplier. No third-party endorsement or independent clinical review is claimed.