New benchmark Psych-ECA evaluates synthetic control arms in psychiatry trials

Researchers have introduced Psych-ECA, a new benchmark designed to evaluate methods for creating synthetic control arms in psychiatric drug development. This benchmark addresses the need for evaluating accuracy, uncertainty calibration, robustness to informative sampling, and false-positive rates in trial decisions. Psych-ECA generates longitudinal symptom trajectories for depression, anxiety, and psychosis, offering known counterfactuals and realistic measurement noise. Initial benchmarking of eight estimators revealed that trajectory-based and flexible machine learning methods achieved the best accuracy, while only the Scribe method demonstrated both accuracy and calibrated uncertainty. AI

IMPACT This benchmark could improve the efficiency and reliability of psychiatric drug development by providing better synthetic control arms.

RANK_REASON The item describes a new benchmark and evaluation of methods for longitudinal psychiatry, fitting the research category. [lever_c_demoted from research: ic=1 ai=0.4]

Read on arXiv cs.LG →

• Ham and Stone
• PHQ-9
• PK-PD Study of Mycophenolic Acid (CellCept) in Pediatric Kidney Transplant Patients
• Positive and Negative Syndrome Scale
• Psych-ECA
• scribe
• Star Drifters

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →