Behavioral Data-Driven Prediction of Suicide Risk Using Machine Learning Approaches

The Suicide Risk Prediction Dataset comprises 1,128 structured records collected from multiple medical institutions and open sources in Bangladesh. This dataset has been curated to support machine learning and statistical modeling for suicide risk assessment and prediction. Each record captures critical demographic, behavioral, and psychological features, enabling effective feature analysis and risk factor identification for suicidal behavior.

Data Collection Details

Collected From:

1.Dhaka Medical College & Hospital – 38 records
2.Bangladesh Medical University (BMU) – 19 records
3.Enam Medical College & Hospital – 27 records
4.Shaheed Tajuddin Ahmed Medical College – 36 records
5.Online Survey (Google Form) – 358 records
6.Open-source data (GitHub) – 650 records

Collection Period: February 2025 – July 2025

Collection Methods:

Hospital records supervised by mental health professionals

Online survey conducted through Google Forms

External open-source datasets carefully selected from GitHub

Dataset Statistics

Total Records: 1,128

Features per Record: 16 structured features + 1 dependent variable

Features Include:

1.Age

2.Gender

3.Religion

4.Occupation

5.Civil Status

6.Level of Education

7.Psychological Session

8.Past Attempt

9.Disorder

10.Illness

11.Alcohol

12.Anger

13.Sleep Problem

14.Isolation

15.Humiliation

16.Sad/Weary

Dependent Variable:

Attempted? (Binary indicator: history of suicide attempt or not)

Ethical Considerations

All data were collected under ethical approval and professional supervision. Mental health experts from BMU, Dhaka Medical College, and other institutions were directly involved to ensure accuracy, patient confidentiality, and clinical relevance.

Key Applications

Risk Factor Analysis: Identifying major behavioral and psychological triggers behind suicidal tendencies.

Predictive Modeling: Training machine learning algorithms for suicide risk prediction.

Healthcare Decision Support: Assisting clinicians in early detection of high-risk patients.

Public Health Policy: Supporting targeted interventions and preventive strategies.

Academic Research: Providing a benchmark dataset for behavioral health and suicide prevention studies.