+ HEALTH Background Discharge letters are essential for continuity of care, yet patients often misunderstand their con…± HEALTH Hematoxylin and Eosin (H&E) glass slide preparation is fundamental to histopathology. Pathologists' expertise…+ CLIMATE Traditional soil heavy metal risk assessments suffer from fragmented indices and poor logical integration. Co…± HEALTH Background Neonatal pneumothorax can progress rapidly, yet detection on supine chest radiographs remains chal…+ HEALTH T cell exhaustion has widespread implications for the progression and treatment of chronic diseases including…+ HEALTH Background Standard pulmonary arterial hypertension (PAH) risk models may not fully apply to PAH associated w…+ HEALTH Objectives To map and synthesize the scientific evidence regarding the use of Augmented Reality (AR) and Mixe…± HEALTH Objective To examine whether the outputs of machine learning algorithms designed to predict risk of cardiovas…
TruaceTracing the truth around AIMonday, September 14, 2026
Subscribe
TRV-2026-1072Certified recordPeer-reviewed

Temporal and cross-site validation of an AI system for self-harm detection

Adequate self-harm surveillance is a key part of suicide prevention. Our previous research demonstrated that an artificial intelligence (AI)-based system could effectively detect self-harm in emergency department triage notes. However, the system was developed using data from a single hospital, raising concerns about its generalisability. Here, we aim to validate the system prospectively and externally to better understand its portability across hospitals. We leveraged emergency department data from two Australi…

Health · The Trace — both readings · certified 2026-09-13 · v1 · article view · machine-readable

Current reading — gain

An AI system combining text normalisation with 1931 features maintained stable self-harm detection in prospective validation at its development metropolitan hospital, achieving PR AUC 0.84 over 329,655 triage notes in the following four years.

Current reading — problem

When applied to a regional hospital 150 km outside Melbourne, the same AI system's ability to distinguish self-harm cases declined to PR AUC 0.78, with instability linked to linguistic domain shift and different self-harm presentations.

What this doesn’t fix

System was developed from a single metropolitan hospital, limiting generalisability, and regional differences in presentation and language reduced stability.

Evidence

Reader signal

How should this claim be treated?

Cite this record

Truvace Impact Record TRV-2026-1072, v1: “Temporal and cross-site validation of an AI system for self-harm detection.” Truvace, 2026-09-13. /record/TRV-2026-1072 (accessed at citation time). sha256 0f67e1fd3ba14571…

Calibration history

Every change to this record since certification, in the open. None yet — the reading has held since it entered the record.

  1. Certifiedv10f67e1fd3ba1…

    Certified into the record

Verify this record
How to verify without trusting this page

Fetch the canonical text of any version from /api/record/TRV-2026-1072 and hash it yourself — for example shasum -a 256 on the saved canonical field. The result must equal content_hash, and each version’s text ends with prev:followed by the prior version’s hash (version 1 chains to 64 zeros). If a single character of any version had been altered since certification, the chain would not reproduce.