ARTIFICIAL INTELLIGENCE IN THE DIAGNOSIS OF OCCUPATIONAL DISEASES: TECHNOLOGICAL INNOVATIONS AND MEDICAL CERTIFICATION IN ASBESTOSIS
Abstract
Background: Asbestosis remains most significant challenges in occupational medicine. The traditional medical certification process, primarily based on the subjective visual assessment of conventional chest radiographs (CXR), is hindered by high interobserver variability. This leads to prolonged administrative disputes and delays in awarding compensation to affected workers. This review critically evaluate the impact of quantitative computed tomography (QCT) and artificial intelligence (AI) on the objectification and standardization of diagnostics and medico-legal certification in asbestosis. Methods: A systematic literature review from 2010 to 2026 was conducted by searching the PubMed, Scopus, Web of Science, and Google Scholar databases. Publications concerning the use of machine learning algorithms, deep learning models, and advanced imaging techniques in occupational lung diseases were analyzed. Results: The application of QCT enables the conversion of interstitial changes into measurable, objective imaging biomarkers. Meanwhile, deep learning algorithms allows for automated, high-throughput detection and staging of pulmonary fibrosis, radically accelerating the evaluation of insurance claims. However, real-world validation studies demonstrate that AI models trained on historical diagnoses can replicate the errors of human experts. Furthermore, the "black box" nature of these algorithms poses ethical and legal challenges, blurring diagnostic accountability. Conclusions: The fusion of AI and QCT represents a breakthrough toward fair and objective medical certification. Nevertheless, due to ethical and cognitive dilemmas, full automation of this process is currently unjustified. The priority remains the development of explainable artificial intelligence systems and the implementation of hybrid models, where automated analysis supports the final decisions of reduced expert panels.