Evolution of Machine Learning-Based Human-Disease Prediction Models: A Bibliometric Analysis of Statistical Learning, Model Validation, and Predictive Performance

Abstract

Background: This study aims to provide a comprehensive bibliometric examination of machine learning applications in disease prediction, focusing on the evolution of machine learning frameworks and predictive performance from 2000 to July 2026 publication records. Using data from the Dimension AI databases, this research maps the field’s intellectual landscape by identifying the most influential authors, organisations, and emerging thematic trends. Method: The study used bibliometric analysis and science mapping of records retrieved from Dimensions AI. The literature identification and screening process was reported using a PRISMA 2020 framework. The dataset was cleaned and analysed using Bibliometrix/Biblioshiny and VOSviewer, with performance analysis, collaboration analysis, title-term analysis, and thematic mapping. Results: Publication growth accelerated sharply and reached its highest level in 2025, while the citation trend showed a notable peak in 2011. The study identified strong international collaboration, with China, the United States, India, and the United Kingdom among major contributors. Recent research has increasingly emphasized disease, patients, healthcare, deep learning, artificial intelligence, validation, and clinical prediction. Conclusion: The research landscape has evolved from algorithm-oriented development toward clinically oriented, inter- pretable, and validated prediction models. Future research should prioritize disease-specific prediction using deep learning together with prospective validation, transparency, fairness, generalisability, and clinical implementation.