Noninvasive Diagnosis of Early-Stage Chronic Kidney Disease and Monitoring of the Hemodialysis Process in Clinical Practice via Exhaled Breath Analysis Using an Ultrasensitive Flexible NH3 Sensor Assisted by Pattern Recognition

Abstract

To achieve the early diagnosis of chronic kidney disease (CKD), noninvasive hemodialysis monitoring, and accurate determination of dialysis duration and adequacy, a noninvasive, point-of-care, user-friendly device should be developed. Here, a flexible, room temperature NH₃ gas sensor sensitive to the key breath biomarkers of CKD─NH₃ and creatinine─was fabricated. The sensor had detection limits of 100 ppb for NH₃ and 1 ppm for creatinine. Clinically, a total of 96 exhaled breath samples, half from 39 CKD patients and the other half from 48 healthy controls were collected and analyzed. With the assistance of a pattern recognition algorithm , the early diagnosis of CKD was achieved by the sensor, with PCA being used due to sensor’s cross-sensitivity to CKD biomarkers. Diagnostic models distinguishing CKD versus non-CKD and early-stage CKD versus advanced-stage CKD were constructed using the SVM algorithm, achieving an overall accuracy of 0.93 and 0.94, with area under the curve (AUC) values of 0.97 and 0.99 for all subjects in receiver operating characteristic (ROC) analysis, respectively. The hemodialysis processes of patients were monitored in real-time, with the sensor response values exhibiting ideal exponential decay over time. The sensor response values showed a strong positive correlation with serum creatinine levels (r = 0.85) and a moderate positive correlation with blood urea nitrogen levels (r = 0.62), both of which are key clinical diagnostic indicators for CKD. These are good results, as 54% of CKD samples are from early-stage CKD patients. These results suggest that the sensor could serve as a noninvasive alternative to traditional blood tests for renal function evaluation and CKD diagnosis. Overall, this sensor demonstrates great potential in clinical practice for early diagnosis of CKD, monitoring the daily health status of CKD patients, optimizing the dialysis schedule, and monitoring the dialysis process in real-time.