Machine Learning-Driven Wearable Sweat Sensors with AgNW/MXene for Non-Invasive SERS-Based Cardiovascular Disease Detection

Traditionally, cardiovascular disease diagnosis has relied on invasive testing methods, which may cause anxiety or discomfort in patients. Noninvasive diagnostic technologies, such as wearable sensors, offer significant advantages by reducing these concerns. Surface-enhanced Raman spectroscopy (SERS) is an ultrasensitive technique well-suited for disease diagnosis, particularly when combined with machine learning (ML) algorithms. In this study, we introduce AgNW/MXene hydroxyl composite membranes as SERS substrates for cholesterol detection in sweat. The high specific surface area of MXene enhances the adsorption of target molecules, thereby improving SERS signal sensitivity. Cholesterol was successfully detected at a concentration of 10 –8 M, demonstrating the robustness of the method over 50 stretch-release cycles. A random forest (RF) model was employed to classify sweat samples from healthy individuals and cardiovascular patients, achieving an accuracy of 83.5%. Compared to traditional invasive methods, our approach provides a noninvasive, highly sensitive, and durable alternative with the added advantage of easy integration into wearable diagnostic devices. The high sensitivity and durability of these wearable sweat sensors highlights their potential for advancing noninvasive cardiovascular disease detection.