A Multidomain Model for Dementia Classification using Harmonized LASI and LASI-DAD Data

A harmonized LASI-DAD Wave 1 logistic regression model classified clinically rated dementia in 2,528 Indian adults with ROC-AUC 0.953 and Brier score 0.037.

• Why it matters: Cognitive test performance in India is confounded by education, literacy, and language, so dementia classification based on cognition alone may miss neurodegeneration-related signal.
• What they did: Using CDR consensus (dementia defined as CDR ≥1; 192 cases, 7.6%), the study compared five algorithms with nested cross-validation and quantified domain contributions via paired bootstrap ROC-AUC differences.
• The result: Informant-reported decline (IQCODE) drove most gains, increasing ROC-AUC by 0.052, while sociodemographic (+0.0020) and cardiometabolic assays (+0.0032) added negligible discrimination.

AI summary of the abstract · may contain errors · not medical advice · read the paper.