Solving the 'Goldilocks problem' in dementia clinical trials with multimodal AI

The development of effective therapeutics for Alzheimer's Disease and related dementias (ADRD) has been hindered by patient heterogeneity and the limitations of current diagnostic tools. New treatments have no chance of working if given to patients who cannot benefit from them. This perspective explores how advances in Artificial Intelligence (AI), particularly multimodal machine learning, can solve the 'Goldilocks problem' of identifying patients for inclusion in clinical trials and support precision treatment in real-world healthcare settings. We examine the challenges of patient stratification, grounded by a conceptual framework of identifying each person's stage and subtype of dementia. We review data from several clinical trials of Alzheimer's disease therapeutics, to explore how AI-guided patient stratification can improve trial outcomes, reduce costs and improve recruitment. Further, we discuss the integration of AI into clinical workflows, the importance of model interpretability and generalizability, and ethical imperative to address algorithmic bias. By combining AI with scientific insight, clinical expertise, and patient experience, we argue that intelligent analytics can accelerate the discovery and delivery of new diagnostics and therapeutics, ultimately transforming dementia care and improving outcomes for patients around the globe.