Post

New research · Psychiatry
JMIR mental health · 1d
Systematic reviewJMIR mental health · 2026

Optimizing Treatment Strategies in the Bipolar Disorder Spectrum With Classical AI Approaches: Systematic Review of Performance, Bias, and Clinical Applicability.

Silvia De Francesco, Damiano Archetti, Cesare Michele Baronio … Alberto Redolfi
Read paper
PsychiatrySystematic review

Biomarker- and cellular-based Artificial Intelligence models predict long-term bipolar disorder treatment response with high accuracy.

Optimizing Treatment Strategies in the Bipolar Disorder Spectrum With Classical AI Approaches: Systematic Review of Performance, Bias, and Clinical Applicability.

Silvia De Francesco et al. · JMIR mental health · 2026
Background

Despite pharmacological advances, treatment response varies significantly among patients, making the identification of individualized treatment strategies a major challenge.

Purpose

The present systematic review aimed to examine the current evidence on classical AI-supported treatment optimization in the BD spectrum.

Methods

The review was conducted in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) 2020 guidelines.

n = 35 studies
96%-99%
Results
biomarker- and cellular-based Artificial Intelligence models correctly predict long-term treatment success
n = 35 studies
More results

Acute symptomatic response models performed modestly (pooled area under the curve [AUC] 0.68), while imaging improved accuracy (74%-77%).

Relapse and readmission prediction achieved a pooled AUC of 0.71, with digital phenotyping and rule-based methods performing best (AUC 0.85-0.88).

Safety and dose optimization models achieved 85%-97% accuracy.

More results

However, 3 studies were considered at high risk of bias due to small sample sizes associated with disproportionately high-performance estimates.

“
Conclusion · 1 of 2

The adoption of classical AI tools in BD serves as a driver for therapeutic optimization, although current AI tools in BD should still be considered exploratory rather than ready for clinical use.

Conclusion · 2 of 2

Effective implementation in real-world clinical scenarios requires more robust, transparent, and externally validated models to ensure reliability and generalizability.

Read paper
0 comments

No comments yet. Be the first.

Related papers

LatestFoundational
AI / Informatics
0.89-0.93
the high level of agreement between the AI rater and psychiatrists on consultation sections
Cohort Study
95.5%
patients identified as low risk by the model did not experience psychiatric worsening
AI / Informatics
0.72
This accuracy placed top LLMs at the 64th percentile of clinician performance.
AI / Informatics
0.705
this is how accurately Claude 4 predicted if someone had PTSD symptoms
Guideline
“To enhance local relevance, countries should adapt recommendations to national service and policy contexts.”
Cross-sectional
approximately 50%
participants reported less severe depression symptoms
Guideline
“Clinicians should not use parenteral pharmacological prophylaxis in adults hospitalised with psychiatric illness at low risk of VTE; and should consider using parenteral pharmacological prophylaxis for high-risk adults with no contraindications.”
Case Report
4 weeks
delusions, paranoia, and thoughts of self-harm went away in this time