AI-Powered Sweat Sensors Capable of Predicting Anxiety Before Symptoms Appear

Aug 8
2 min read

A person is wearing a white medical sensor on an arm band, attached to their bicep. The sensor has a small drop icon and an indicator light.

Revolutionary innovations in wearable technology are transforming health monitoring, utilizing artificial intelligence (AI) to analyze sweat and offer early, personalized diagnoses. These advancements promise to change how we detect and manage medical conditions, from anxiety to cancer.

A recent study introduced Stressomic, an AI-powered wearable microfluidic biosensor capable of predicting anxiety before symptoms appear. This innovative skin patch measures cortisol (CORT), epinephrine (EPI), and norepinephrine (NE) in sweat, with sequential sampling approximately every six minutes. The device uses a flexible printed circuit board (FPCB) with gold nanodendrite (AuND)-decorated laser-engraved graphene (LEG) electrodes, and incorporates temperature, pH, and ionic strength sensors. Random forest (RF) models trained with initial hormone data predicted negative affect, positive affect, and state anxiety with accuracies of 62%, 54%, and 86%, respectively. This technology is capable of distinguishing physical exertion from psychological strain, paving the way for personalized stress dashboards and early detection of maladaptive responses.

In an equally promising area, a hydrogel-based wearable sweat sensor combining Surface-Enhanced Raman Spectroscopy (SERS) with AI has been developed to monitor the treatment effect of lung cancer. This non-invasive, portable, and high-frequency approach is crucial for more comfortable oncology treatment. Using multiple AI algorithms, such as LGB, GNB, LDA, RF, and SVM, the diagnostic model achieved an accuracy of 89.7% in classifying three treatment effects (progressive disease, partial response, and no change) based on over 12,000 SERS spectra from clinical patients. Clinical data revealed that carbonyl biomarkers in sweat might be crucial for understanding complications like diabetes and hypertension.

Furthermore, researchers from CNR-ISOF and Ca’ Foscari University of Venice have integrated AI with 2D-3D composite materials to detect and quantify ions like sodium (Na⁺) and potassium (K⁺) in biological fluids such as sweat and saliva. A notable advance is that this system can employ non-optimized or poorly selective sensors, outperforming traditional methods thanks to sophisticated deep learning trained with over four million samples. This capability is fundamental for real-time biofluid monitoring and analysis in complex environments.

Together, these studies highlight the immense potential of AI-powered sweat sensors to revolutionize personalized medicine, offering non-invasive diagnostics and real-time monitoring of various health conditions, from mental health to chronic disease treatment.