“Trying to Do No Harm”: exploring clinician concerns towards the use of AI for risk prediction in psychiatry
Bibliographic record
Abstract
Background: The application of artificial intelligence (AI) in healthcare is expanding, including in psychiatry. However, its successful adoption depends on clinician acceptance and trust. Despite the growing interest, there remains a knowledge gap in understanding the clinician perspectives and concerns, towards AI in psychiatry. Objective: This qualitative, pre-implementation study explored clinician concerns and perceived barriers towards the application of predictive AI for clinical outcomes in a large mental health hospital. Methods and Analysis: Four virtual focus groups were conducted with 16 clinicians who provided care at a large mental health hospital in Ontario, Canada. Two focus groups (n=9) included physicians, and two (n=7) included allied clinicians. Participants discussed their awareness and concerns with predictive AI for clinical outcomes. Transcripts were analysed using reflexive thematic analysis. Results: Six themes emerged regarding clinician willingness to use and implement AI for clinical outcome prediction in mental healthcare: AI model performance, quality of data sources, system issues, end-user behaviours, patient outcomes and clinician well-being. Subthemes included the absence of technical infrastructure, quality data to support AI development, the 'black box phenomenon' of AI algorithms, loss of critical thinking, medicolegal concerns and the potential harms from over-intervening. Conclusion: To our knowledge, this is the first qualitative study that uses focus groups to explore a full range of clinician attitudes towards machine learning-based prediction tools in mental healthcare settings. It highlights major areas of concern for emerging AI technology. Understanding clinicians' perspectives is critical to identifying barriers to the introduction of AI in psychiatry.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.055 | 0.127 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.009 | 0.018 |
| Scholarly communication | 0.006 | 0.007 |
| Open science | 0.003 | 0.008 |
| Research integrity | 0.004 | 0.007 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".