Machine Learning Enabled Portable Electrical Impedance Tomography System for Community Level Chronic Kidney Disease...

Chronic kidney disease (CKD) is a prevalent and progressive condition that affects 850 million people around the world and poses significant healthcare challenges due to its high morbidity, late detection, and costly treatment. Current diagnostic methods, such as serum creatinine-based eGFR and urinary albumin tests, are invasive, expensive, and require laboratory settings, limiting accessibility for community- and home-based remote monitoring. This study proposes a novel, portable, and non-invasive approach for CKD screening and eGFR estimation using electrical impedance tomography (EIT) with machine learning. A total of 138 subjects, including healthy volunteers and CKD patients, were assessed using a portable EIT system. Renal EIT signals were obtained across a range of frequencies and processed using principal component analysis (PCA) to extract relevant features. A two-stage machine learning model was developed, utilizing support vector machines (SVM) for CKD classification and XGBoost for eGFR prediction. Patient-specific anthropometric parameters, such as age, sex, height, weight, and waist circumference, were integrated into the model to enhance accuracy. Models were trained with five-fold cross-validation and evaluated on separate testing sets. The proposed system achieved an overall classification accuracy of 93% (sensitivity: 100%; specificity: 94%) in distinguishing healthy individuals (Stage 1) from CKD patients (Stages 2-4) with an AUC-ROC of 0.97. For eGFR prediction, the regression model achieved an R² value of 0.57 (p < 0.001), showing significant differences between CKD stages. Compared to anthropometric-based models, the proposed approach demonstrated superior performance in both classification and regression tasks. This study demonstrates the feasibility of a low- cost, portable EIT system combined with machine learning for point-of-care CKD detection and eGFR estimation. The system has the potential to enable widespread, non-invasive, and self- administrable CKD screening at a community level and telemedicine settings, reducing healthcare costs and improving early detection and management of CKD.Clinical Relevance-Chronic kidney disease (CKD) is a silent, progressive condition and a major contributor to global morbidity and mortality, with eGFR being a critical diagnostic metric. A portable EIT system enables low-cost, non-invasive CKD screening and eGFR estimation, addressing the urgent unmet healthcare need for accessible early detection and monitoring in home and community settings.