CPSA · Spoken practice

Talk to a patient

Everything else in Finalist trains recognition — read a framework, rate yourself, pick an option. The CPSA tests production: opening your mouth and generating the consultation under a clock. Speak to the simulated patient, or type if you'd rather; the station is scored the same either way.

Chest pain — focused history

Mr Doyle, 58, has come to the emergency department with chest pain that started this morning. Take a focused history. You do not need to examine him.

Gerald Doyle · 7 min · 8 station-critical of 16 cues

Breathlessness — focused history

Mrs Iqbal, 71, has been referred by her GP with breathlessness that has worsened over the past fortnight. Take a focused history.

Nasreen Iqbal · 7 min · 8 station-critical of 15 cues

Breaking bad news — a new cancer diagnosis

Mr Whitfield, 64, had a CT scan last week for weight loss and abdominal pain. It shows a pancreatic mass with liver metastases. He has come for the result, alone. The MDT has not yet met. Explain the result to him.

Alan Whitfield · 8 min · 10 station-critical of 13 cues

Medication counselling — starting warfarin

Mrs Barnes, 68, has newly diagnosed atrial fibrillation and a mechanical mitral valve. She is being started on warfarin. Counsel her about the medication.

Joan Barnes · 7 min · 10 station-critical of 14 cues

The angry relative — a delayed diagnosis

Mr Hollis's mother was admitted four days ago with abdominal pain. A CT yesterday showed a perforation that was not identified on her admission imaging. He has asked to speak to a doctor and is visibly angry. Speak with him.

Dean Hollis · 8 min · 10 station-critical of 13 cues

What this does and doesn't do

The patient is an authored script, not an AI — every line they can say is written down and clinically checked, so they never invent a symptom that contradicts the case. Across 5 stations there are 71 markable cues, 46 of them station-critical.

Scoring is coverage: did you ask? That is the honest limit of what a transcript supports. It does not score rapport, empathy or fluency — grading warmth from speech-to-text would be a guess dressed up as feedback, and you would rightly stop trusting the rest of the app.

Speech recognition needs Chrome or Edge and your microphone; nothing is recorded or uploaded by Finalist — the transcript stays on your device and only the score reaches your profile. Typed mode works everywhere.