Evidence map›Paper›PMID 41270691›Full record

ArticlePsychiatry research2026

Integrating expert knowledge into large language models improves performance for psychiatric reasoning and diagnosis.

Karthik V Sarma, Kaitlin E Hanss, Andrew J M Halls, Andrew Krystal, Daniel F Becker, Anne L Glowinski, Atul J Butte

Abstract read
PubMed Full text at PMC Publisher
In one paragraph

Article in Psychiatry research, 2026. The graph could read no effect estimate from its abstract, so it casts no vote on the map. Cited by 4 papers.

0numbers the graph read from it
0cells of the map it votes in
4citing papers in PubMed
–field-weighted citation impact
1 · What the graph read from it

What it found

Each row is one number read from the abstract, on the scale the paper reported it, with its interval. Left of the dashed line favours the treatment, right favours the comparator. Under each row is the sentence it came from. New to these charts? A ten-minute tutorial.

The abstract states no effect estimate the extractor could read, or names no intervention and outcome on the map, so this paper lights no cell and moves no belief. It is still indexed, cited and linked below.

2 · The registry

The trial behind it

Trials whose registry record cites this paper, or whose number appears in the abstract. A trial that started after this paper was published is citing it as background, not reporting it.

Neither the registry nor the abstract names a trial number. If this is a trial report, that itself is worth knowing.

3 · Its place in the literature

Who cites it

4 citing papers in PubMed.

  1. Review
  2. Review
  3. Article
  4. Article
4 · The record

Corrections and comments

5 · Who and what money

Authors and funding

7 authors.

Karthik V SarmaDepartment of Psychiatry and Behavioral Sciences, University of California San Francisco, 675 18th Street, San Francisco, CA 94143, USA; Bakar Computational Health Sciences Institute, University of California San Francisco, 550 16th Street, San Francisco, CA 94143, USA. Electronic address: karthik.sarma@ucsf.edu.
Kaitlin E HanssDepartment of Psychiatry and Behavioral Sciences, University of California San Francisco, 675 18th Street, San Francisco, CA 94143, USA.
Andrew J M HallsDepartment of Psychiatry and Behavioral Sciences, University of California San Francisco, 675 18th Street, San Francisco, CA 94143, USA.
Andrew KrystalDepartment of Psychiatry and Behavioral Sciences, University of California San Francisco, 675 18th Street, San Francisco, CA 94143, USA.
Daniel F BeckerDepartment of Psychiatry and Behavioral Sciences, University of California San Francisco, 675 18th Street, San Francisco, CA 94143, USA.
Anne L GlowinskiDepartment of Psychiatry and Behavioral Sciences, University of California San Francisco, 675 18th Street, San Francisco, CA 94143, USA.
Atul J ButteBakar Computational Health Sciences Institute, University of California San Francisco, 550 16th Street, San Francisco, CA 94143, USA.

Funding

TRAINING THE NEXT GENERATION OF MENTAL HEALTH RESEARCHERSR25MH060482 · NIMH · UNIVERSITY OF CALIFORNIA, SAN FRANCISCO · PI DANIEL H MATHALON, Susan M. Voglmaier · 2000 to 2026
$5.2M
Research Colloquium for Junior InvestigatorsR13DA042568 · NIDA · AMERICAN PSYCHIATRIC ASSOC FOUNDATION · PI CLARKE, DIANA ELAINE · 2017 to 2019
$75k
Research Colloquium for Junior Psychiatrist InvestigatorsR13DA052195 · NIDA · AMERICAN PSYCHIATRIC ASSOC FOUNDATION · PI CLARKE, DIANA ELAINE · 2020 to 2022
$75k
Research Colloquium for Junior Psychiatrist InvestigatorsR13DA060048 · NIDA · AMERICAN PSYCHIATRIC ASSOC FOUNDATION · PI Diana Elaine Clarke · 2024 to 2026
$75k
NIDA NIH HHS R13 DA042568NIDA NIH HHS R13 DA052195NIDA NIH HHS R13 DA060048NIMH NIH HHS R25 MH060482
6 · The paper itself

Abstract

BACKGROUND AND

methodsThe authors sought to evaluate the performance of common large language models (LLMs) in psychiatric diagnosis, and the impact of integrating expert-derived reasoning on their performance. Clinical case vignettes and associated diagnoses were retrieved from the DSM-5-TR Clinical Cases book. Diagnostic decision trees were retrieved from the DSM-5-TR Handbook of Differential Diagnosis and refined for LLM use. Three LLMs were prompted to provide diagnosis candidates for the vignettes either by directly prompting or using the decision trees. These candidates and diagnostic categories were compared against the correct diagnoses. The positive predictive value (PPV), sensitivity, and F

resultsWhen directly prompted to predict diagnoses, the best LLM by F DISCUSSION: When used to predict psychiatric diagnoses from case vignettes, direct prompting of the LLMs yielded most true positive diagnoses but had significant overdiagnosis. Integrating expert-derived reasoning into the process using decision trees improved LLM performance (as measured by F

Indexed as

Clinical ReasoningLanguageMental DisordersDecision TreesDiagnosis, DifferentialDiagnostic and Statistical Manual of Mental DisordersHumansLarge Language ModelsArtificial intelligenceDecision supportDiagnosisDSMGenerative artificial intelligencePsychiatryReasoning

Identifiers

PMID41270691
PMCPMC12883249

What OpenQuestion holds

Textmetadata
LicenceTDM
Read underepoch 390

Registered trials

None linked

Read under generation 80e0d062 · epoch 390. Bibliography from PubMed, PubMed Central and OpenAlex; grants from NIH RePORTER; trial links from ClinicalTrials.gov; estimates, votes and beliefs from the OpenQuestion graph.