Abstract Personalized Treatment Recommendation Systems (PTRS) are increasingly central to precision medicine, enabling data-driven selection of therapies tailored to individual patient characteristics, including clinical history, biological markers, and behavioral context. Despite rapid methodological advances, existing research remains fragmented, with limited consensus on evaluation practices and insufficient consideration of real-world deployment constraints. This survey provides a systematic and critical review of 60 peer-reviewed studies published between 2019 and 2024, selected using a PRISMA guided protocol from major scholarly databases. The reviewed literature is organized into a unified taxonomy encompassing content-based, collaborative filtering, knowledge-based, and hybrid treatment recommendation paradigms, with emphasis on their clinical applicability and algorithmic design. Recent advances in deep learning and reinforcement learning are examined for their role in improving adaptability and treatment personalization. To enable meaningful cross-study comparison, this work introduces a multidimensional benchmarking rubric that evaluates PTRS beyond predictive performance, incorporating criteria such as algorithmic rigor, clinical relevance, validation depth, data transparency, scalability, and explainability. The analysis reveals persistent challenges related to model interpretability, population generalizability, data privacy, and ethical compliance, which continue to hinder clinical adoption. Emerging solutions, including federated learning, explainable AI, and privacy-preserving architectures, are discussed as promising pathways toward trustworthy deployment. By synthesizing methodological trends, practical limitations, and evaluation gaps, this survey offers a structured foundation and forward-looking perspective for the development of robust, transparent, and clinically viable PTRS in modern healthcare.
Rathod et al. (Mon,) studied this question.