Predictive Models and Risk Scoring System Development for Assessing Survival in Uterine Cancer Patients With Type 2 Diabetes Mellitus: A Territory-Wide Cohort Study
Author(s)
Chenwen Zhong, PhD1, Junjie Huang, PhD2, Martin Chi Sang Wong, MD2;
1The Chinese University of Hong Kong, Postdoctoral fellow, Hong Kong, China, 2The Chinese University of Hong Kong, Hong Kong, Hong Kong
1The Chinese University of Hong Kong, Postdoctoral fellow, Hong Kong, China, 2The Chinese University of Hong Kong, Hong Kong, Hong Kong
OBJECTIVES: To develop predictive models and establish a risk-scoring system that identifies factors associated with survival in uterine cancer patients with type 2 diabetes mellitus(T2D) and estimates their survival probabilities.
METHODS: A population-based, retrospective cohort study of uterine cancer patients with T2D managed in the Hong Kong public health care system between 2000 and 2020 was conducted. Five machine learning algorithms were utilized to develop predictive models for survival: Cox proportional hazards regression, survival tree, LASSO Cox regression, boosting, and random survival forest(RSF). The performance of each model was evaluated using a time-dependent area under the curve(AUC) and concordance index(C-index). Key factors were identified through Shapley Additive Explanations analysis based on the top-performing model, while the AutoScore-Survival package facilitated the development of a risk-scoring system.
RESULTS: This cohort study included 2,047 uterine cancer patients with T2D, with an average survival time of 100.82 months(standard deviation 72.75). The RSF model exhibited the strongest predictive performance, achieving a time-dependent AUC of 0.823 and a C-index of 0.90. A risk-scoring system was created based on several criteria: age at cancer diagnosis, duration of T2D, creatinine levels, serum potassium levels, LDL cholesterol levels, body mass index (BMI), and triglyceride levels. This scoring system classified 31.4% of patients as high-risk, with a 5-year survival probability of 43.5%, approximately 1.7 times lower than that of the low-risk group.
CONCLUSIONS: These findings underscore the significance of the risk scoring system and highlight the roles of age at cancer diagnosis, duration of T2D, creatinine levels, serum potassium levels, LDL cholesterol levels, BMI, and triglyceride levels as critical factors in identifying high-risk uterine cancer patients with T2D. By incorporating these factors into clinical assessments, healthcare professionals can better tailor interventions to enhance survival outcomes. Future research should focus on validating the risk-scoring system across diverse populations and further exploring the underlying mechanisms.
METHODS: A population-based, retrospective cohort study of uterine cancer patients with T2D managed in the Hong Kong public health care system between 2000 and 2020 was conducted. Five machine learning algorithms were utilized to develop predictive models for survival: Cox proportional hazards regression, survival tree, LASSO Cox regression, boosting, and random survival forest(RSF). The performance of each model was evaluated using a time-dependent area under the curve(AUC) and concordance index(C-index). Key factors were identified through Shapley Additive Explanations analysis based on the top-performing model, while the AutoScore-Survival package facilitated the development of a risk-scoring system.
RESULTS: This cohort study included 2,047 uterine cancer patients with T2D, with an average survival time of 100.82 months(standard deviation 72.75). The RSF model exhibited the strongest predictive performance, achieving a time-dependent AUC of 0.823 and a C-index of 0.90. A risk-scoring system was created based on several criteria: age at cancer diagnosis, duration of T2D, creatinine levels, serum potassium levels, LDL cholesterol levels, body mass index (BMI), and triglyceride levels. This scoring system classified 31.4% of patients as high-risk, with a 5-year survival probability of 43.5%, approximately 1.7 times lower than that of the low-risk group.
CONCLUSIONS: These findings underscore the significance of the risk scoring system and highlight the roles of age at cancer diagnosis, duration of T2D, creatinine levels, serum potassium levels, LDL cholesterol levels, BMI, and triglyceride levels as critical factors in identifying high-risk uterine cancer patients with T2D. By incorporating these factors into clinical assessments, healthcare professionals can better tailor interventions to enhance survival outcomes. Future research should focus on validating the risk-scoring system across diverse populations and further exploring the underlying mechanisms.
Conference/Value in Health Info
2025-05, ISPOR 2025, Montréal, Quebec, CA
Value in Health, Volume 28, Issue S1
Code
RWD42
Topic
Real World Data & Information Systems
Disease
SDC: Diabetes/Endocrine/Metabolic Disorders (including obesity), SDC: Oncology