A retrospective Chinese study reported that a CT-based multimodal deep learning model could distinguish pulmonary Talaromyces marneffei infection from pulmonary tuberculosis in HIV/AIDS patients. The approach uses radiomics and deep learning features together and was evaluated on a performance set with reported AUC values and specificity metrics. The authors reported an AUC of 0.931, with a specificity result of 0.981 for differentiating the two infectious entities—an important diagnostic challenge because symptoms and imaging can overlap, particularly in immunocompromised populations. While still early and retrospective, the reported discrimination performance suggests a potential decision-support role in settings where rapid microbiologic confirmation may be delayed or unavailable. Prospective validation would be needed before any clinical workflow integration.