A hierarchical machine learning model for predicting self-harm and suicidal behavior in hospitalized patients with schizophrenia using clinical history and nursing observations - Report - MDSpire

A hierarchical machine learning model for predicting self-harm and suicidal behavior in hospitalized patients with schizophrenia using clinical history and nursing observations

  • By

  • Liang, Chen

  • Meng, Xianfeng

  • Duan, Ying

  • Yang, Wei

  • Zhu, Gang

  • Wang, Jinhuan

  • Sun, Ying

  • Wang, Mingtao

  • Liu, Miao

  • Sun, Chenhui

  • Hu, Kunyuan

  • Shao, Wei

  • Ren, Jintao

  • Shao, Xiaojun

  • Zhang, Yang

  • April 24, 2026

  • 0 min

Share

Clinical Report: Two-Tiered Machine Learning for Self-Harm in Schizophrenia

Overview

This study developed a two-layered machine learning framework to identify schizophrenia inpatients at high risk for self-harm and suicidal actions. The model effectively integrates clinical data and nursing assessments, achieving high predictive accuracy.

Background

Suicide and self-harm are critical concerns in psychiatric care, particularly among individuals with schizophrenia. Identifying patients at high risk is essential for timely intervention and prevention strategies. This study addresses the need for improved risk assessment tools in inpatient settings to enhance patient safety.

Data Highlights

ModelAUCSensitivitySpecificityAccuracy
Static LR Model0.7564---
Dynamic SVM Model0.8531---
Combined Hierarchical Model0.90480.85420.77890.8042
Flat Combined-Feature Model0.90220.6667--

Key Findings

  • Previous self-harm is a strong predictor of future self-harm/suicidal behavior (OR = 4.323).
  • Hopelessness/depression significantly increases risk (OR = 3.090).
  • Younger age is associated with a lower risk (OR = 0.938).
  • Dynamic indicators like negative self-evaluation and self-reported depression are critical for risk assessment.
  • The hierarchical model outperformed traditional flat models in sensitivity and overall predictive accuracy.

Clinical Implications

The findings suggest that integrating clinical history with ongoing nursing assessments can enhance the identification of high-risk patients in psychiatric settings. Clinicians should consider implementing such machine learning frameworks to support individualized care and timely interventions.

Conclusion

This study highlights the potential of a hierarchical machine learning approach to improve risk detection for self-harm and suicidal behavior in schizophrenia inpatients, ultimately aiding in the development of preventive strategies.

Related Resources & Content

  1. Frontiers in Psychiatry, 2026 -- Development and validation of a machine learning–based risk prediction model for non-suicidal self-injury in adolescents
  2. BMC Psychiatry, 2025 -- Modeling Predictive Factors for Suicidal Thoughts in Individuals Experiencing Cognitive Decline
  3. BMC Psychiatry, 2025 -- Development of a machine learning model for predicting compulsory psychiatric care using clinical notes
  4. BMC Psychiatry, 2026 -- Creation and assessment of a machine learning framework for detecting individuals at elevated risk for psychotic disorders through analysis of medical records
  5. BMC Psychiatry, 2026 -- Time to suicide after psychiatric inpatient discharge: a nationwide Swedish survival analysis
  6. VA/DoD CLINICAL PRACTICE, 2024 -- Clinical Practice Guideline for Suicide Risk
  7. Machine learning algorithms and their predictive accuracy for suicide and self-harm: Systematic review and meta-analysis
  8. Time to suicide after psychiatric inpatient discharge: a nationwide Swedish survival analysis | BMC Psychiatry | Springer Nature Link
  9. VA/DoD CLINICAL PRACTICE

Original Source(s)

A hierarchical machine learning model for predicting self-harm and suicidal behavior in hospitalized patients with schizophrenia using clinical history and nursing observations

  • by Liang, Chen , Meng, Xianfeng , Duan, Ying , Yang, Wei , Zhu, Gang , Wang, Jinhuan , Sun, Ying , Wang, Mingtao , Liu, Miao , Sun, Chenhui , Hu, Kunyuan , Shao, Wei , Ren, Jintao , Shao, Xiaojun , Zhang, Yang

  • April 24, 2026

Related Content

Exploring the Influence of Genetic Vulnerability to Neurodevelopmental Disorders on Antipsychotic Efficacy in Schizophrenia: Insights from a Postmortem Brain Analysis

Romanian male patients with the dual diagnosis of schizophrenia and alcohol use disorder: a prospective study of clinical, social, and treatment-related factors affecting quality of life

FDA Approves Auvelity for Agitation in Alzheimer’s Disease

The oral combination therapy expands treatment options for agitation associated with dementia due to Alzheimer’s disease.