AIによる認知症の早期診断

抄録

With the rapid increase in the number of dementia cases worldwide and the development of disease–modifying therapy with anti–amyloid antibodies for Alzheimer disease, the demand for a more concise and precise method for diagnosing dementia at earlier stages has risen. The use of artificial intelligence (AI) has been gaining ground in medicine, particularly in dementia research―it has become an important tool for analyzing patient–derived data. AI–driven analyses of brain images, biofluid biomarkers, and genetic information could increase the accuracy of dementia diagnosis. For example, a machine learning model applied to magnetic resonance imaging to diagnose early–stage dementia reportedly achieves high accuracy with over 80% sensitivity and specificity. Furthermore, AI–driven analysis can reveal information on patient–derived data that may not be accessible via conventional approaches. Thus, AI–powered analysis could lead to a deeper understanding of the pathogenesis of dementia and the development of new targets for dementia therapies. AI–assisted methods for dementia screening have been proposed and tested in real–world settings in combination with cutting–edge sensing techniques. Low explainability has been a major bottleneck in utilizing AI in medical diagnosis ; however, new techniques, such as Shapley Additive exPlanations (SHAP), which are used to explain the individual predictions made by the AI model, could help in overcoming issue. In this review article, recent progress in AI–related research in the dementia field is discussed along with the challenges and benefits of AI use in this field.