A RESEARCH team led by Macao University of Science and Technology (MUST) has developed a new AI-based clinical diagnostic aid model called IRENE, which processes multimodal input in a unified manner. IRENE is reportedly the first of its kind and designed to jointly learn holistic representations of medical images, unstructured chief complaints, and structured clinical information to facilitate medical decision-making. The study, which involved researchers from the West China Hospital of Sichuan University and the University of Hong Kong, has been published in the latest edition of the peer-reviewed scientific journal Nature Biomedical Engineering. The increasing demand for precision medicine coupled with limited medical resources in some regions has made machine-learning techniques the de facto choice for automatic intelligent medical diagnosis. Deep learning has enabled machine-learning models to detect diseases from medical images almost as good as human doctors. Nevertheless, interpreting medical images in their clinical context and coming to a precise diagnosis remains a challenge for the AI model. Professor Zhang Kang, the leader of this study, claimed that IRENE has the ability to interpret multimodal clinical information, using intra- and intermodal attentional methods consistent with daily clinical practices. IRENE was applied to pulmonary disease identification and clinical outcome prediction in patients infected with COVID-19, yielding desirable results, according to Zhang.(Xinhua) | 