Comparing Epidemiological Sampling Techniques: From Classical Non-Probability Methods to AI-Enhanced Recruitment Strategies

Abstract

Background: This review article compares traditional sampling techniques in epidemiology, particularly non-probability methods, with emerging artificial intelligence (AI)-enhanced recruitment strategies. Classical methods such as convenience, purposive, snowball, and quota sampling remain widely used due to their feasibility in resource-constrained and hard-to-reach settings, despite inherent risks of selection bias and limited generalizability.

Methods: By synthesizing evidence from recent methodological and epidemiological studies, this review examines the strengths, limitations, ethical implications, and practical applications of both traditional non-probability approaches and AI-driven strategies.

Results: AI-driven approaches leverage machine learning, natural language processing (NLP), predictive analytics, and digital data sources to optimize participant identification, reduce recruitment inefficiencies, and potentially mitigate certain forms of bias under appropriate data and governance conditions. The findings suggest that AI is transforming epidemiological recruitment not by replacing classical methods, but by augmenting them through adaptive, data-driven hybrid frameworks.

Conclusion: Responsible integration of AI, guided by epidemiological principles and ethical safeguards, is essential to maximize public health impact. Future research should focus on hybrid models that balance the feasibility of classical methods with the precision of AI.