Optimizing Patient Recruitment for Clinical Trials: A Hybrid Classification Model and Game-Theoretic Approach for Strategic Interaction

Abstract

This research is imperative due to the pressing need for improved patient recruitment in clinical trials, addressing challenges such as delays and high costs. By introducing a classification model and a game theoretic approach for clinical trial setting, we aim to boost trial efficiency, advance healthcare research, and enhance patient outcomes. This research is critical for revolutionizing recruitment strategies and accelerating medical progress. In this paper, we present a classification model that has been specifically designed to address this issue effectively. The proposed model employs an Autoencoder, augmented by a super classification model that merges Logistic Regression, Support Vector Machines, Random Forest Trees, and Decision Trees using a stacking classifier. The output of the super classifier is further processed by a meta classifier to obtain the final result. Notably, the model achieves a training accuracy of 99.576% and a validation accuracy of 83.45%, illustrating its robust classification performance and its potential to streamline patient recruitment, reducing delays and resource consumption. In addition to the classification model, this study formulates a three-layer game theoretic model involving Patients, Doctors or Clinical Investigators, and Research Firms. Within this static repeated game setting, players sequentially strategize to optimize their recruitment strategies, while research firms aim to optimize their overall interaction. The paper proposes a novel optimal solution that strikingly balances the payoffs of all three players. Moreover, the work presents a necessary condition and closed form for the existence of an equilibrium in the game, offering a strategic approach to recruitment optimization, and striking a balance between stakeholders. This equilibrium-seeking solution has the potential to revolutionize recruitment dynamics and foster collaboration. Additionally, the study’s theoretical contributions lay the groundwork for future research in this critical healthcare domain.

Keywords

• Clinical trials
• decision-making
• ensemble learning
• feature extraction
• game theory
• healthcare analytics