August 9 – August 14, 2026
Gold-standard clinical diagnosis of dementia and cognitive impairment is often not feasible in large population-based cohort studies because it is time- and resource-intensive for participants and research teams. Algorithmic classification is an efficient alternative that can leverage existing data to establish a proxy for cognitive impairment and dementia without requiring resource-intensive clinical evaluation. This year’s conference will explore best practices for developing accurate classification algorithms that can be used by researchers to answer questions about the prevalence, progression, and determinants of dementia in cognitive aging studies. This multidisciplinary workshop will provide hands-on training through Workgroups using real-world datasets.
This multidisciplinary workshop will provide hands-on training through workgroups using real-world datasets. Workgroups topics will include (1) comparing associations of biomarkers (e.g., MRI metrics) with different forms of dementia classifications including clinical, algorithmic classification, and machine learning-based predictions; (2) feature engineering for machine learning classification to address what components should be considered best practice for inclusion in classification algorithms; (3) approaches to develop and validate a dementia algorithm in a sample where one does not have a reference standard diagnosis; and (4) development of probability-based algorithms (e.g., Bayesian) that incorporate population incidence/prevalence rates and empirical assessment data.