Generative Artificial Intelligence-Aided Image-Based Tuberculosis Diagnosis

Abstract

Generative artificial intelligence systems such as large language models are coaxed via prompt engineering into generating suggestions for the development of an automated system for the diagnosis of tuberculosis on the basis of chest radiography image sequences. The recommendations of the generative artificial intelligence are followed through to construct artificial intelligence models and then these models are trained, tested and validated on suitably formatted data and harnessed for the automated detection of tuberculosis via the analysis or processing of chest radiography images. The performance of the trained artificial intelligence models could be enhanced with a view to fielding them in modules for the automated image-based diagnosis of tuberculosis as part of a comprehensive artificial intelligence-powered healthcare system that could provide clinical decision support to medical doctors and healthcare professionals.