ISL/ACHS eTóraxLaboral AI Pneumoconiosis Detection System (Deep Learning Chest X-ray Analysis)

Overview

Chile's Instituto de Seguridad Laboral (ISL) and the Asociación Chilena de Seguridad (ACHS), in collaboration with the technology company Convicción Digital, developed and deployed eTóraxLaboral, an artificial intelligence system for detecting pneumoconiosis — a group of incurable occupational lung diseases including silicosis and asbestosis — from chest X-ray radiographs. The system uses deep learning image analysis to provide diagnostic decision support for occupational health physicians within Chile's occupational disease surveillance infrastructure.

The development of eTóraxLaboral was motivated by the need to improve timely screening for workers exposed to crystalline silica dust, particularly in mining, construction, and related industries. According to Chile's Ministry of Health report on occupational silica exposure, an estimated 5.4 percent of the Chilean workforce is at high risk of exposure. Pneumoconiosis diagnosis requires interpretation of chest radiographs according to the International Labour Organization (ILO) classification system, which demands specialised training. eTóraxLaboral was designed to support this workflow by helping medical teams identify radiographs that may show pneumoconiotic opacities.

eTóraxLaboral is an intelligent platform that analyses digital chest radiographs to detect signs of pneumoconiosis using deep learning algorithms trained on labelled radiographic datasets. The system is designed as a diagnostic support tool — it assists medical teams in making evidence-based decisions by flagging radiographs that exhibit features consistent with pneumoconiotic opacities, rather than replacing the clinical judgement of qualified physicians. The platform is compatible with HL7 (Health Level 7) and DICOM (Digital Imaging and Communications in Medicine) standards, enabling integration with existing hospital and clinic radiology information systems and picture archiving and communication systems (PACS).

A clinical validation study, published in the Journal of Occupational and Environmental Medicine (April 2025), evaluated eTóraxLaboral's diagnostic performance through a retrospective analysis of 2,300 randomly selected chest radiographs. The study assessed sensitivity, specificity, false positive and negative rates, positive and negative predictive values, likelihood ratios, efficiency, error rate, and the area under the receiver operating characteristic (ROC) curve using a Fagan nomogram. The reported results showed a positive likelihood ratio (LR+) of 23 and a negative likelihood ratio (LR-) of 0.2, indicating strong discriminative performance in the study setting. The authors also noted some false positives linked to anatomical overlap and increased lung markings, and less frequent false negatives involving misinterpretation of pneumoconiotic opacities as consolidation-type findings. The study was conducted by Eduardo R. Ruiz, Carolina A. Arellano, Carmen A. Archila, Carolina Llobet, Gonzalo Carrasco, and Francisca Pinochet, with affiliations to Convicción Digital (Talca, Chile) and Mutual de Seguridad (Santiago, Chile).

The International Social Security Association (ISSA) recognised eTóraxLaboral as a Good Practice (gp/244268) for using artificial intelligence to detect pneumoconiosis, describing it as an AI system for detecting respiratory pathologies of occupational origin. The ISSA recognition highlights the system's role in supporting earlier detection and treatment of occupational respiratory disease.

From a social protection perspective, eTóraxLaboral operates within Chile's occupational health and safety system governed by Law 16.744, which establishes mandatory social insurance against occupational accidents and diseases. The ISL is the public entity responsible for administering this social insurance for workers not covered by private mutual insurance associations (mutuales). The ACHS is one of three private mutual associations providing occupational accident and disease insurance in Chile. The system supports occupational disease surveillance and early detection workflows that can inform treatment and compensation processes.

The system's broader significance lies in its potential relevance for other countries seeking to strengthen occupational lung disease screening capacity. Chile's earlier validation of digital radiography for pneumoconiosis screening by the Instituto de Salud Pública (ISP) in 2012 provided an enabling regulatory context for AI-assisted analysis of digitised chest X-rays.

Convicción Digital, the technology company that developed eTóraxLaboral, is a teleradiology company based in Talca, Chile, focused on artificial intelligence and occupational medicine applications for respiratory disease diagnosis. The platform is available as a cloud-based service for healthcare providers and occupational health clinics.

Classification

Programme Details

AI-assisted pneumoconiosis detection system deployed within Chile's occupational health insurance framework (Law 16.744) by ISL and ACHS, using deep learning analysis of chest X-rays to support occupational health physicians in early diagnosis of silicosis and other pneumoconioses among workers exposed to crystalline silica.

Implementation Details

Risk & Oversight

Deployment & Outcomes

Sources

1. SRC-003-CHL-001 Convicción Digital (n.d.) 'eTórax — Convicción Digital', Convicción Digital. Available at: https://www.prod-convicciondigital.cl/ (Accessed: 26 March 2026).
   https://www.convicciondigital.cl/
2. SRC-001-CHL-001 Ruiz, E.R., Arellano, C.A., Archila, C.A., Llobet, C., Carrasco, G. and Pinochet, F. (2025) 'Clinical Validation of an AI System for Pneumoconiosis Detection Using Chest X-rays', Journal of Occupational and Environmental Medicine, 67(4), pp. e250–e256. doi: 10.1097/JOM.0000000000003329.
   https://pubmed.ncbi.nlm.nih.gov/39905932/
3. SRC-002-CHL-001 ISSA (n.d.) 'Using artificial intelligence to detect pneumoconiosis — An AI system for detecting respiratory pathologies of occupational origin', International Social Security Association Good Practice gp/244268. Available at: https://www.issa.int/gp/244268 (Accessed: 26 March 2026).
   https://www.issa.int/gp/244268