Explainable machine learning integrating bioelectrical impedance for 6-month cardiovascular risk in peritoneal...

Area of Science:

• Nephrology
• Cardiology
• Artificial Intelligence
• Medical Informatics

Background:

• Cardiovascular disease (CVD) is a leading cause of morbidity in patients undergoing peritoneal dialysis (PD) for end-stage renal disease.
• Inadequate fluid management in PD patients is a significant risk factor for CVD.
• Bioelectrical impedance spectroscopy (BIS) is a valuable tool for assessing fluid status in PD patients.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI) model for predicting major adverse cardiovascular events (MACEs) within six months in clinically stable PD patients.
• To integrate bioelectrical impedance spectroscopy (BIS) measurements with patient medical history for enhanced cardiovascular risk prediction.
• To evaluate the performance of different AI algorithms in predicting MACEs in this patient population.

Main Methods:

• Development of an AI model using logistic regression, random forest, XGBoost, and deep neural networks.
• Training and testing data split (80% training, 20% testing) with synthetic minority over-sampling technique for class imbalance.
• Model performance evaluation using area under the ROC curve (AUC), calibration plots, and decision curve analysis (DCA); feature ablation studies were conducted.

Main Results:

• The random forest model demonstrated the highest performance with an AUC of 0.88.
• Cardiovascular disease (CVD) history was identified as the most influential predictor.
• Integrating BIS features significantly improved diagnostic sensitivity from 0.68 to 0.84, highlighting the value of fluid status assessment.

Conclusions:

• A 15-feature random forest AI model, incorporating BIS measurements and medical history, accurately predicts 6-month MACE risk in stable PD patients.
• The model shows strong discriminatory ability and potential clinical utility as confirmed by DCA.
• Combining BIS data with medical history offers a significant improvement in predicting cardiovascular events for PD patients.