An explainable framework for dementia threat prediction primarily based on way of life information utilizing machine studying and deep studying | Uncover Artificial Intelligence

Dementia is a progressive neurodegenerative illness that impacts tens of millions worldwide and is a problem for early detection as a result of price, invasiveness, and restricted entry of obtainable scientific strategies. This examine proposes an explainable Artificial Intelligence (AI) system for the prediction of threat for dementia primarily based solely on way of life parameters, thus offering a non-invasive and economical possibility. We examined a spectrum of machine studying strategies (Decision Tree, Random Forest, Okay-Nearest Neighbor, Naïve Bayes, Support Vector Machine, and Multilayer Perceptron) and likewise deep studying fashions (Hybrid Neural Network). Out of these, Random Forest exhibited one of the best accuracy of 93.1% on way of life information and 95.2% on an unbiased exterior validation dataset, exhibiting glorious generalizability. To fight the essential challenge of mannequin explainability, SHapley Additive exPlanations (SHAP) was utilized for offering each the worldwide and native interpretability and highlighting the significance of modifiable way of life parameters, e.g., bodily exercise, smoking, and comorbidities, within the willpower of dementia threat. Class imbalance was overcome utilizing strategies like SMOTE, Gaussian augmentation, and scaling, thus enabling dependable detection of minority lessons. Results verify that the mixture of explainable AI and environment friendly ML/DL strategies supplies correct, interpretable, and clinically important prediction of dementia threat. The constructed system has the potential for supporting lively healthcare interventions, particularly in resource-constrained environments the place superior analysis tools isn’t accessible.