Ai-based prediction models for suicide risk assessment and prevention: a cross-sectional study.

Sethy Rati; Dash Aniket; Pany Sanjaya

doi:10.51168/tcmq1q40

Authors

Rati Ranjan Sethy Associate Professor, Department Of Psychiatry, Government Medical College And Hospital, Phulbani Odisha, India
Dash Aniket Assistant Professor, Department of Psychiatry, Pabitra Mohan Pradhan Medical College & Hospital, Talcher, Odisha, India
Sanjaya Kumar Pany Associate Professor, Department of Psychiatry, Hi Tech Medical College, Bhubaneswar, Odisha, India

DOI:

https://doi.org/10.51168/tcmq1q40

Keywords:

Artificial intelligence, Machine learning, Mental health, Prediction model, Psychiatry, Suicide risk

Abstract

Background

Suicide risk is influenced by interacting clinical, psychological, and psychosocial factors. This study evaluated the utility of an artificial intelligence (AI)-based prediction model for suicide risk assessment among psychiatric patients.

Aim

To evaluate the predictive utility of an AI-based model and identify factors associated with higher suicide risk.

Materials and Methods

A hospital-based cross-sectional study was conducted among 60 adult psychiatric patients attending Government Medical College and Hospital, Phulbani, Odisha. Demographic, clinical, and psychosocial data were collected using a structured proforma. Suicide risk was assessed using standardized clinical assessment. An AI-based prediction model was developed using selected predictors and evaluated using 5-fold cross-validation. Logistic regression and diagnostic performance measures were used for analysis.

Results

Depression (p=0.002), substance use disorder (p=0.018), previous suicide attempt (p<0.001), and family history of suicide (p=0.021) were significantly associated with suicide risk. Recent stressful life events (p=0.004), poor social support (p<0.001), financial difficulties (p=0.048), and family conflict (p=0.002) were associated with moderate-to-high risk. Previous suicide attempt (adjusted OR 5.82; p=0.007), poor social support (adjusted OR 4.95; p=0.010), depression (adjusted OR 4.31; p=0.018), and substance abuse (adjusted OR 3.68; p=0.037) were independent predictors. The AI model achieved 86.7% accuracy, 88.2% sensitivity, 84.0% specificity, AUC-ROC 0.91, and 84.5% cross-validation accuracy.

Conclusion

AI-based prediction may support early identification of patients at increased suicide risk alongside comprehensive psychiatric assessment.

Recommendations

Larger multicentre studies with external validation and assessment of real-world clinical impact are recommended before widespread implementation.

References

1. Wang N, Luo F, Shivtare Y, Badal VD, Subbalakshmi KP, Chandramouli R, et al. Learning models for suicide prediction from social media posts. Proc CLPsych. 2021.

https://doi.org/10.18653/v1/2021.clpsych-1.9

2. Belsher BE, Smolenski DJ, Pruitt LD, Bush NE, Beech EH, Workman DE, et al. Prediction models for suicide attempts and deaths: a systematic review and simulation. JAMA Psychiatry. 2021.

3. Lejeune A, de Timary P, Noirhomme Q, et al. Machine learning and the prediction of suicide in psychiatric populations: a systematic review. Neurosci Biobehav Rev. 2024.

4. Collins GS, Moons KGM, Dhiman P, et al. TRIPOD+AI statement: updated guidance for reporting clinical prediction models that use regression or machine learning methods. BMJ. 2024;385: e078378.

5. Rana R, Higgins N, Haque KN, et al. AI-assisted summary of suicide risk formulation. arXiv. 2024.

6. Kim KH, Kim JH, Kim MG, Han HW. Real-world implementation and predictive performance of suicide prediction models: a systematic review and meta-analysis. Digit Health. 2026.

https://doi.org/10.1177/20552076261456585

7. Elbarbary K, Alfudoul A, Samir AA, et al. Are artificial intelligence chatbots safe for suicide risk assessment? A narratively synthesized review of current evidence. Glob Ment Health (Camb). 2026.

https://doi.org/10.1017/gmh.2026.10260

8. Izmaylov D, Segal A, Gal K, et al. Combining domain knowledge with large language models for predicting suicide risk in online counselling services. Nat Lang Process. 2026.

https://doi.org/10.1017/nlp.2026.10022

9. Jacobucci R, Shao W, Kobrinsky V, Ammerman B. Predicting momentary suicidal ideation from smartphone screenshots using vision-language models: prospective machine learning study. JMIR Ment Health. 2026.

https://doi.org/10.2196/90581

10. Artificial intelligence in suicide risk assessment: a systematic literature review. Discover Artificial Intelligence. 2026.

11. American Psychiatric Association. Practice guideline for the assessment and treatment of patients with suicidal behaviors: updated recommendations. Washington, DC: APA; 2024.ss