How do current inclusion/exclusion criteria, diagnostic requirements, and biomarker thresholds in CNS trials contribute to high screen failure rates, and where can we safely relax or refine them without compromising data integrity?

To what extent do disease and patient heterogeneity (genetic, phenotypic, comorbidities, concomitant meds) dilute treatment signals in CNS trials and how can stratification, enrichment, and adaptive designs be more systematically embedded upstream? What are the potential disadvantages of using these methods as far as generalizability of results, and how is the appropriate balance determined?

Are we seeing high early attrition in CNS programs primarily because of flaws in signal-seeking study design (sample size, endpoint selection, visit schedule) rather than true lack of efficacy, and how can early-phase designs be recalibrated to de-risk later stages?

How much of screen and post-randomization attrition is driven by operational friction—site capabilities, rater burden, visit intensity, travel and caregiving demands—and what proven tactics best reduce dropout in cognitively or functionally impaired populations?

Where can sponsors, sites, and regulators better align on pragmatic, patient-centered CNS trial frameworks (e.g., decentralized assessments, flexible visit windows, real-world data integration) to lower screen failure, manage heterogeneity, and still meet evidentiary standards for approval?