Related Experiment Video
Updated: Aug 14, 2026
12:18
A Machine Learning Approach to Design an Efficient Selective Screening of Mild Cognitive Impairment
Published on: January 11, 2020
Nurse-Led Identification of Financial Toxicity in Stroke Patients Using Machine Learning: Development and Validation
Yuan Song1, Yunjing Xing1, Ce Zong2
1School of Nursing and Health, Zhengzhou University, Zhengzhou, Henan, China, zzu.edu.cn.
Journal of Nursing Management
|July 21, 2026
Summary
This study developed an XGBoost model to predict financial toxicity (FT) in stroke patients, enabling nurses to identify those at risk. The tool aids in providing timely support and resources for financially vulnerable individuals.
Area of Science:
- Health Informatics
- Machine Learning in Healthcare
- Oncology Nursing
Background:
- Financial toxicity (FT) is a growing concern for patients with chronic conditions, including stroke survivors.
- Limited evidence exists on FT prevalence and identification in stroke patients.
- Early FT identification by nurses can facilitate crucial financial, psychosocial, and discharge support.
Purpose of the Study:
- To develop and validate a machine learning (ML) model for nurse-led identification of FT in stroke patients.
- To create a practical tool for assessing individual FT risk in clinical settings.
Main Methods:
- A cohort of 575 stroke patients was analyzed using Health Ecology Theory principles.
- The Least Absolute Shrinkage and Selection Operator (LASSO) method was used for feature selection.
- Five ML models were trained, with XGBoost selected as optimal and validated externally. SHapley Additive exPlanations (SHAP) were used for interpretation.
Main Results:
- The prevalence of FT among stroke patients was 62.3%.
- The XGBoost model achieved high performance (AUC 0.823 internal, 0.865 external validation).
- Key FT predictors included age, fear of progression, complications, caregiver status, and out-of-pocket costs.
Conclusions:
- The validated XGBoost model demonstrates strong performance for identifying FT in stroke patients.
- A web-based tool derived from the model can support nurses in real-time FT assessment.
- Further longitudinal validation is recommended to confirm clinical utility and integration into nursing workflows.