Advances in risk prediction models for cancer-related cognitive impairment.

Abstract

Cancer-related cognitive impairment (CRCI) has emerged as a significant long-term complication in cancer survivors, particularly those undergoing chemotherapy, radiotherapy, or targeted therapies. Despite advances in treatment, CRCI affects patients' quality of life, impacting their daily functioning, work capacity, and psychological well-being. In recent years, research has focused on identifying predictive factors for CRCI and developing risk prediction models to facilitate early intervention. This review summarizes the latest progress in CRCI risk prediction models, including traditional statistical approaches such as logistic regression and advanced machine learning techniques. While machine learning models demonstrate superior predictive performance, limitations such as data availability and model interpretability remain. Additionally, the review highlights key risk factors-such as age, cancer type, and treatment modalities-and evaluates the strengths and weaknesses of various predictive models in terms of accuracy, generalizability, and clinical applicability. Finally, this paper discusses the challenges in validating these models across diverse populations and the need for further research to enhance model reliability and personalization of interventions.