Key result
A novel model-agnostic approach applied to an XGBoost model identified ischemic heart disease as the most influential risk factor (effect score 1.75) for developing heart failure in patients with type 2 diabetes.
Why the study?
Complex machine learning models using EHR data lack transparency, requiring a model-agnostic approach to interpret risk factor effects on developing heart failure in patients with type 2 diabetes.
Does the proposed model-agnostic approach improve the accuracy of ranking risk factors for developing heart failure in patients with type 2 diabetes compared to built-in XGBoost importance scores?
Observational (n=723)
Yes
Does the proposed model-agnostic approach improve the accuracy of ranking risk factors for developing heart failure in patients with type 2 diabetes compared to built-in XGBoost importance scores?
A novel model-agnostic approach improves the accuracy of ranking risk factors for heart failure in patients with type 2 diabetes compared to standard XGBoost importance scores.
Hypothesis-generating for ischemic heart disease prioritization in type 2 diabetes; leaves open prospective validation of the model-agnostic approach.
Objective: Understanding the risk factors for developing heart failure among patients with type 2 diabetes can contribute to preventing deterioration of quality of life for those persons. Electronic health records (EHR) provide an opportunity to use sophisticated machine learning models to understand and compare the effect of different risk factors for developing HF. As the complexity of the model increases, however, the transparency of the model often decreases. To interpret the results, we aimed to develop a model-agnostic approach to shed light on complex models and interpret the effect of features on developing heart failure. Using the HealthFacts EHR database of the Cerner EHR, we extracted the records of 723 patients with type 2 diabetes (with at least 6 yeas of follow up), and 134 of them developed heart failure. Using age and comorbidities as the features and heart failure as the outcome, we trained logistic regression, random forest, XGBoost, neural network, and then applied our proposed approach to rank the effect of each factor on developing heart failure. Results: Comparing with “importance score” built-in function from XGBoost, our proposed approach was more accurate in ranking the effect of the different risk factors on developing heart failure.
No takes yet. Share an insight, caveat, or question.
Miran et al. (2021) conducted an observational in Type 2 diabetes and Heart failure (n=723). Model-agnostic effect score approach (XGBoost) vs. Built-in importance score was evaluated on Effect score of risk factors for developing heart failure. A novel model-agnostic approach applied to an XGBoost model identified ischemic heart disease as the most influential risk factor (effect score 1.75) for developing heart failure in patients with type 2 diabetes.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: