Precision Mining of Potential Recruited Patients for Clinical Trials Based on Multimodal AI Algorithms

DOI: https://doi.org/10.62517/jike.202604301

Author(s)

Nanyang Guo, Chen Gong, Peng Shen

Affiliation(s)

Waterdrop Research Institute, Beijing, China

Abstract

The recruitment of clinical trial patients faces challenges such as information heterogeneity, semantic gap, and lack of temporal correlation. Traditional methods are difficult to achieve accurate and efficient potential subject mining. This article proposes a precision mining framework based on multimodal AI algorithms. Firstly, it elaborates on the medical informatics foundation of multimodal data fusion, semantic representation of inclusion and exclusion standards, and unified representation theory of heterogeneous patient data; Furthermore, key challenges such as difficulty in aligning multimodal data, mismatched information granularity, bottlenecks in text image fusion, and missing temporal window associations will be analyzed; Finally, a multimodal semantic alignment method for inclusion criteria, a joint representation learning architecture that integrates text and images, a dynamic matching model for temporal perception, and an interpretability mining framework are proposed. This framework provides a systematic technical path and theoretical support for intelligent clinical trial patient screening.

Keywords

Multimodal AI Algorithm; Clinical Trials; Accurate Patient Excavation; Semantic Alignment of Input and Output Standards

References

[1] Liu H Y, Yang Y, Li G L, et al. Application of Python-based Apriori Algorithm in Information Mining of Patients with Extremely Long Hospital Stay. Chinese Medical Record, 2022, 23(06): 68-71. [2] Wang Q H. Research on Risk Factor Mining of Plateau-Related Cholelithiasis Patients Based on Machine Learning Intelligent Algorithms. Qinghai Province: Qinghai Normal University, 2021-06-18. [3] Jin H P, Tao Y Q, Li Z H, et al. Survival Prediction Algorithm for Gastric Cancer Patients Based on Multimodal Multi-Instance Learning. Journal of Computer-Aided Design & Computer Graphics, 2025, 37(02): 349-360. [4] Park E S, Wollstein Y, Hutchison L A, et al. Recruitment of Minority Patients in Metabolic Dysfunction-Associated Steatotic Liver Disease Clinical Trials. Digestive Diseases and Sciences, 2026 (prepublish): 1-7. [5] Tan Y H, Peng J Y, Fang C, et al. Design and Application of an Intelligent Recruitment and Screening Information System for Trial Participants. Tumor Pharmacy, 2025, 15(04): 475-479.