Development and Evaluation of an Optimal Machine Learning Model for Predicting Nutritional Risk in Nasopharyngeal
Benxiang Zhu1, Chang Gao1, Peijuan Chen2
1School of Nursing, Southern Medical University, Guangzhou, China.
Aim:
To develop a predictive model for nutritional risk in patients with nasopharyngeal carcinoma (NPC) and identify clinically meaningful ranges for key risk factors to guide early intervention.
Methods:
This study enrolled 520 patients with nasopharyngeal carcinoma (NPC) who underwent radiotherapy at Guangzhou Nanfang Hospital from 2021 to 2024. Thirty-two baseline variables were collected, including body measurements, lab tests, treatment details, and lifestyle factors. Seven machine learning models were developed. Key predictors were selected using LASSO regression, and their importance was assessed using SHAP values.
Results:
The XGBoost model performed best, with an AUC of 0.775 on the validation set. Four main predictors of nutritional risk were identified: body mass index (BMI), alanine transaminase (ALT), clinical stage, and smoking status. Patients with a BMI between 21.5 and 24.9 kg/m2 and ALT values in the higher range of normal had a lower risk of malnutrition. These findings provide more specific guidance than existing tools.
Conclusion:
This study highlights the added value of combining clinical data and machine learning to identify both key predictors and their optimal ranges for nutritional risk.
More Related Videos
07:15Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
04:09Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
Published on: October 10, 2018