Forecasting Survival in Nasopharyngeal Cancer Patients With Diabetes Through Machine Learning: A Population-Based Analysis
Author(s)
Junjie Huang, PhD1, Chenwen Zhong, PhD2, Martin Chi Sang Wong, MD1.
1The Chinese University of Hong Kong, Hong Kong, Hong Kong, 2The Chinese University of Hong Kong, Hong Kong, China.
1The Chinese University of Hong Kong, Hong Kong, Hong Kong, 2The Chinese University of Hong Kong, Hong Kong, China.
OBJECTIVES: The objective of this study is to assess the risk factors affecting the survival of nasopharynx cancer patients with type 2 diabetes mellitus (T2DM) and create a risk score to predict the survival probability of patients.
METHODS: We collected data from Hong Kong Hospital Authority Data Collaboration Laboratory (HADCL) including 1,132 nasopharynx cancer patients with T2DM between 2000 and 2020. The mean survival time (SD) of the target subjects was 88.05 (71.65) months, 268 (23.67%) were female, 356 (31.45%) were dead. Four algorithms including CoxPH, CoxNet random survival forest (RSF), and Survival tree were utilized. Area under the curve (AUC) and concordance index (C-index) were used to compare their performances. SHAP (Shapley Additive Explanations) analysis was performed for risk factors identification and model outputs attribution.
RESULTS: Significant predictors for nasopharynx cancer survival among T2DM patients included cancer diagnosis age (HR=1.07, 95%CI [1.06, 1.08], p<.001), DM duration (HR=1.08, 95%CI [1.04,1.11], p<.001), anti-lipid use (HR=0.61, 95%CI [0.48, 0.77], p<.001), alcohol (HR=1.38, 95%CI [1.08, 1.76], p=0.011), sex (HR=1.48, 95%CI [1.08, 2.03], p=0.014), CHD (HR=1.44, 95%CI [1.02, 2.04], p=0.039), and BMI (HR=0.96, 95%CI [0.93, 1.00], p=0.041). The RSF model have the best predicting performance (AUC 0.866, c-index 0.85). The tuned model of risk scores derived through AutoScore-Survival were assigned based on the following criteria: diagnosis age of cancer less than 50 years (0), 50-60 years (15), 60-70 years (29), over 70 years (42); DM duration period < 1 year (0), 1-5 years (33), more than 5 years (38); HbA1c <7 (0), ≥7 (3); HDL_c < 1.0 (0), HbA1c ≥ 1.0 (1); Anti-lipid = 1 (0), =0 (12).
CONCLUSIONS: This study identified that diagnosis age, DM duration, anti-lipid use, alcohol, sex, CHD, and BMI were associated with prognosis of nasopharynx cancer patients with T2DM. RSF model has the best predicting power.
METHODS: We collected data from Hong Kong Hospital Authority Data Collaboration Laboratory (HADCL) including 1,132 nasopharynx cancer patients with T2DM between 2000 and 2020. The mean survival time (SD) of the target subjects was 88.05 (71.65) months, 268 (23.67%) were female, 356 (31.45%) were dead. Four algorithms including CoxPH, CoxNet random survival forest (RSF), and Survival tree were utilized. Area under the curve (AUC) and concordance index (C-index) were used to compare their performances. SHAP (Shapley Additive Explanations) analysis was performed for risk factors identification and model outputs attribution.
RESULTS: Significant predictors for nasopharynx cancer survival among T2DM patients included cancer diagnosis age (HR=1.07, 95%CI [1.06, 1.08], p<.001), DM duration (HR=1.08, 95%CI [1.04,1.11], p<.001), anti-lipid use (HR=0.61, 95%CI [0.48, 0.77], p<.001), alcohol (HR=1.38, 95%CI [1.08, 1.76], p=0.011), sex (HR=1.48, 95%CI [1.08, 2.03], p=0.014), CHD (HR=1.44, 95%CI [1.02, 2.04], p=0.039), and BMI (HR=0.96, 95%CI [0.93, 1.00], p=0.041). The RSF model have the best predicting performance (AUC 0.866, c-index 0.85). The tuned model of risk scores derived through AutoScore-Survival were assigned based on the following criteria: diagnosis age of cancer less than 50 years (0), 50-60 years (15), 60-70 years (29), over 70 years (42); DM duration period < 1 year (0), 1-5 years (33), more than 5 years (38); HbA1c <7 (0), ≥7 (3); HDL_c < 1.0 (0), HbA1c ≥ 1.0 (1); Anti-lipid = 1 (0), =0 (12).
CONCLUSIONS: This study identified that diagnosis age, DM duration, anti-lipid use, alcohol, sex, CHD, and BMI were associated with prognosis of nasopharynx cancer patients with T2DM. RSF model has the best predicting power.
Conference/Value in Health Info
2025-09, ISPOR Real-World Evidence Summit 2025, Tokyo, Japan
Value in Health Regional, Volume 49S (September 2025)
Code
RWD253
Topic Subcategory
Health & Insurance Records Systems
Disease
SDC: Oncology