Supplementary file 3_Can machine learning predict non-suicidal self-injury? A systematic r

A systematic review and meta-analysis of 12 studies involving 33,366 participants, published on figshare in 2026. The document synthesizes the performance of machine learning models for predicting non-suicidal self-injury, reporting pooled metrics such as area under the curve, sensitivity, and specificity. It was authored by Qianhui Wen and is licensed under CC-BY-4.0.

Use Cases

• Benchmarking machine learning model performance for NSSI prediction based on the reported pooled AUC, sensitivity, and specificity metrics.
• Identifying research gaps in the application of deep learning for NSSI prediction based on the limited evidence mentioned.
• Assessing risk of bias in existing NSSI prediction studies based on the application of the PROBAST+AI tool described.
• Informing the design of prospective multicenter studies for NSSI prediction based on the conclusions regarding heterogeneity and validation needs.

Strengths

• Includes a quantitative synthesis of 12 studies with 33,366 total participants.
• Reports specific pooled performance metrics, such as an AUC of 0.83 for ensemble models and 0.75 across all 19 models.
• Employs a standardized risk-of-bias assessment tool (PROBAST+AI) and provides a systematic review registration identifier.

Limitations

• The primary data file is a 113.8 KB DOCX document, which is a small-scale summary report rather than a raw dataset.
• Column-level documentation and sample data are unavailable; the content is a review article.
• The data reflects findings up to June 28, 2025; its utility for tracking the latest model developments may be limited over time.

Provenance

Source

figshare

Collection Method

Systematic review and meta-analysis of studies from multiple databases.

Time Range

Studies from database inception to June 28, 2025.

Freshness

Last updated 2026-04-30 14:01:21.

Geography

null