Leveraging Feature Transfer to Predict Medication Resistance and Secondary-Clinical Outcomes in Psychotic Disorders in Forensic Settings
Bibliographic record
Abstract
Abstract Medication resistance in psychotic disorders represents a critical challenge in forensic psychiatry, where up to 50% of patients show poor treatment response, leading to increased risk of relapse, violence, and rehospitalization. Feature Transfer, a novel machine learning framework based on rank aggregated feature selection, transfers predictive features identified for one outcome to related outcomes while maintaining clinical interpretability, a critical advantage over conventional transfer learning approaches that obscure feature level insights by transferring complex model parameters. Applied to psychotic disorders, this methodology identified key predictors for medication resistance and assessed their transferability to related clinical outcomes. Analyzing data from 893 patients across 11 forensic psychiatric institutions, we compared Feature Transfer models (using the top 25 features discriminating medication resistance from responders) with full feature models (95 features) for predicting clinical relapse, treatment non adherence, and escape behaviors. In the broader psychotic disorders sample, Feature Transfer achieved statistically equivalent performance to full feature models for clinical relapse and treatment non adherence (F1 score differences with confidence intervals overlapping zero), though performed less effectively for escape behaviors (AUC: 0.736 vs 0.838). In schizophrenia patients (n=634), Feature Transfer showed statistically significant improvement in F1 score for clinical relapse prediction compared to full feature models (difference: 0.119, 95% CI: 0.025 to 0.213), with notably higher sensitivity (0.912 vs 0.802) while maintaining comparable discriminative ability (AUC: 0.912 vs 0.925, difference not statistically significant). Treatment history features, particularly previous medication unresponsiveness and duration of clinical care, maintained high predictive importance across multiple clinical outcomes (relapse, non adherence, and escape behaviors), suggesting they represent fundamental risk indicators regardless of the specific outcome being predicted. While our retrospective design limits causal inference and relies on historical indicators as proxies for secondary outcomes (relapse and escape behaviors), the demonstrated utility of medication resistance features across different clinical outcomes reveals potential shared risk dimensions in psychotic disorders, particularly for relapse prediction in schizophrenia. Feature Transfer offers a transparent approach for identifying common predictive factors that could advance personalized intervention strategies in complex psychiatric populations.
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 | 0.005 | 0.013 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.002 | 0.001 |
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".