Pain prediction model based on machine learning and SHAP values for elders with dementia in Taiwan
Shwu-Feng Tsay1, Cheng-Yu Chang2, Sing Shueh Hung3
1Department of Nursing and Health Care, Ministry of Health and Welfare, No. 488, Sec. 6, Zhongxiao E. Rd., Nangang District, Taipei City 115, Taiwan; School of Nursing, National Taiwan University, No.1, Sec. 1, Jen Ai Rd, Taipei City 100, Taiwan; Department of Health Services Administration, College of Public Health, China Medical University, No. 100, Sec. 1, Jingmao Rd., Beitun Dist., Taichung City 406, Taiwan.
International Journal of Medical Informatics
|May 14, 2024
Summary
Machine learning accurately predicts pain in elderly dementia patients using factors like the Karnofsky scale and arthritis. This aids in better pain management for this vulnerable population.
Area of Science:
- Gerontology
- Artificial Intelligence
- Pain Management
Background:
- Pain is prevalent in elderly individuals with dementia, often underrecognized due to cognitive impairment.
- Dementia symptoms can complicate accurate pain assessment and effective management strategies.
Purpose of the Study:
- To develop a machine learning model for predicting the pain index in elderly individuals with dementia.
- To identify key variables influencing pain perception in this population.
Main Methods:
- Utilized questionnaire data from 113 cases to train and compare three machine learning algorithms.
- Employed SHapley additive explanations (SHAP) for model interpretability, identifying feature importance and relationships.
- Implemented feature selection to optimize the prediction model.
Main Results:
- Random forests with feature selection demonstrated superior performance in predicting pain index, minimizing root mean square error and mean absolute error.
- Identified 11 key features, with the Karnofsky scale showing a significant positive association with pain index.
- Arthritis was identified as the most influential disease predictor for pain index in elderly dementia patients.
Conclusions:
- The study provides crucial insights for predicting pain in elderly patients with dementia.
- The developed model has the potential for application in digital tools (e.g., apps, webpages) to enhance efficiency and reduce labor in pain assessment.
