The AI-Guided Clinical Trial Architect: A Genetic Algorithm and MCDM Platform for Adaptive, Multi-Objective Patient Cohort Selection and Trial Simulation

Introduction: Clinical trial design faces a critical challenge in balancing multiple, often conflicting, objectives such as statistical power, patient safety, cohort diversity, recruitment speed, and cost. While Multi-Objective Genetic Algorithms (MOGA) and Multi-Criteria Decision-Making (MCDM) have been applied independently in pharmaceutical contexts, their synergistic potential for clinical trial architecture remains under-explored. This paper introduces the AI-Guided Clinical Trial Architect (AI-CTA), a novel computational platform that integrates a MOGA with a fuzzy MCDM framework for adaptive, multi-objective patient cohort selection and trial simulation. Methods: The methodology involves a multi-phase workflow: data encoding via entropy-based weighting, evolutionary exploration of cohort configurations using a fuzzy-enhanced NSGA-II, and final selection through a Fuzzy TOPSIS analysis that incorporates expert-derived linguistic weights to handle uncertainty. Results & Discussion: A comprehensive case study for an oncology trial (n=200 from 1,850 candidates) demonstrates the platform's efficacy. The MOGA successfully generated a Pareto-optimal set of cohorts, from which the FMCDM module identified an optimal cohort achieving a superior balance of objectives (Closeness Coefficient: 0.656), validating the platform's ability to derive non-intuitive, robust solutions. Conclusion: By unifying the explorative power of MOGA with the deliberative precision of FMCDM, the AI-CTA provides a transformative, transparent, and computationally robust environment for designing more efficient, equitable, and economically viable clinical trials.