Work

Risk prediction models for fear of cancer recurrence in patients with cancer: a systematic review

Year: 2026

Type: review

Abstract: Background: The apprehension of cancer recurrence among cancer survivors strongly influences their quality of life. Predictive models can enable the timely recognition of this concern and the implementation of interventions. In recent years, a multitude of models have been constructed to predict the likelihood of disease recurrence in cancer patients. Nevertheless, the methodological strictness and clinical applicability of these models necessitate further elucidation. Objectives: Systematic assessment of the existing predictive models for fear of cancer recurrence. Methods: FOUR databases, including PubMed, the Cochrane Library, Embase, and Web of Science, were systematically searched to identify relevant studies published from the inception of the databases to December 31, 2025. Data extraction was independently performed by two reviewers on the basis of the critical evaluation and data extraction checklist of the systematic review predictive model study. The quality of the included models was assessed using the predictive model bias risk assessment tool. Results: A total of 1,137 articles were retrieved, from which 15 models across 9 studies were ultimately selected for inclusion. Logistic Regression and machine learning were the most frequently employed methods for model development. The AUC values of the developed models ranged from 0.66 to 0.9786, whereas those reported in validation studies ranged from 0.83 to 0.949. None of the models underwent external validation. All risk prediction models exhibited high bias, primarily because of inappropriate data sources and methodological shortcomings. Social support, age, and education level were identified as common predictive factors. Conclusions: Current predictive models for the risk of cancer recurrence exhibit good discriminative power. However, most of the included studies are single-center retrospective studies with small sample sizes and lack external validation; these studies exhibit significant methodological limitations that restrict their clinical applicability. Future researchers are encouraged to develop predictive models in accordance with the PROBAST guidelines, enhance methodological rigor, and conduct both internal and external validation. Systematic Review Registration: https://www.crd.york.ac.uk/PROSPERO/recorddashboard, identifier CRD420261288735.

Source: Frontiers in Psychiatry

DOI: 10.3389/fpsyt.2026.1940840