Research · Study S012026
AI ambient scribing in routine psychiatric care
Evaluating the use of an artificial intelligence ambient scribe in a Swiss psychiatric clinic: efficiency, documentation quality and therapeutic alliance. A nonrandomized controlled trial at Psychiatrie St. Gallen, led by me as principal investigator.
In short
Ambient scribes listen to the conversation between clinician and patient and draft the clinical note. They promise to save time. But psychiatry is built on conversation: the record, the clinician's skills and the patient's trust may all be affected. This study tests what actually changes when clinicians start using an ambient scribe in everyday care.
For researchers
Clinicians using the AlpineAI Listener Tool are compared with clinicians documenting as usual, before and after introduction. Analysis follows difference-in-differences logic with mixed-effects models. Documentation quality is rated by blinded raters. Full methods are in the protocol.
Study design
From intervention to analysis
Ambient AI scribing in routine outpatient psychiatry
Eight Psychiatrie St. Gallen sites · 47 clinicians
AlpineAI Listener Tool
Usual documentation, no study scribe
Before and after introduction, both groups
Clinician documentation time
Blinded PDQI-9 ratings
Content-coverage scoring
Alliance: WAI-SR
Empathy: CARE Measure
Difference-in-differences logic · mixed-effects models
Secondary: clinician burnout (BAT)
What we measure
| Outcome | Measure | Reported by |
|---|---|---|
| Efficiency | Documentation time | Routine system data |
| Documentation quality | PDQI-9 rating of initial assessment reports; content-coverage scoring | Blinded raters |
| Therapeutic alliance | WAI-SR | Patients |
| Empathy | CARE Measure | Patients |
| Clinician burnout | Burnout Assessment Tool (BAT) | Clinicians |
Timeline
- Aug 2026Intervention phase begins
- Feb 2027Data collection ends
- 2027Three results papers: efficiency, documentation quality, therapeutic relationship
Registration & how to cite
Dobelsek M. Evaluating the Use of an Artificial Intelligence Ambient Scribe in a Swiss Psychiatric Clinic — Efficiency, Documentation Quality, and Therapeutic Alliance: Protocol for a Nonrandomized Controlled Trial. JMIR Preprints. 2026;111815.
Peer-reviewed version in revision at JMIR Research Protocols · DOI to followEthics: the Ethics Committee of Eastern Switzerland (EKOS) confirmed that the study falls outside the scope of the Swiss Human Research Act.
Interested in the study?
I'm glad to discuss the design, share materials or explore collaboration on ambient AI in mental health care.