L'ÉVALUATION DES PERFORMANCES DIAGNOSTIQUES DE DEUX MODÈLES D'INTELLIGENCE ARTIFICIELLE (MAIA AI ET GPT-5.2) DANS L'INTERPRÉTATION DE LA RADIOGRAPHIE THORACIQUE : ÉTUDE COMPARATIVE À L'HÔPITAL MIXTE DE LAGHOUAT

Abstract

Artificial intelligence is playing an increasingly important role in medical imaging, particularly in the interpretation of chest radiographs. The aim of this study was to evaluate and compare the diagnostic performance of two artificial intelligence models, GPT-5.2 and MAIA AI, in identifying pneumonia, pneumothorax, pleural effusion, and normal chest X-rays. This was an analytical observational study with mixed retrospective and prospective data collection, including 110 chest radiographs. The results from both models were compared to the interpretation of an expert pulmonologist, which was considered the gold standard. MAIA AI demonstrated superior diagnostic performance for pneumonia and pneumothorax, while both models showed comparable performance in detecting pleural effusion and normal chest radiographs. These findings suggest that artificial intelligence represents a promising tool for aiding radiological diagnosis, but it should be used to complement human expertise.