AI-Enhanced Flexible IL-6 Sensor—Smart Detection for Early Disease Alerts

Wearable devices have become essential tools in our daily lives, supporting activities from human-machine interfaces to health monitoring. In healthcare, sweat-based wearable sensors enable continuous, minimally invasive tracking of biomarkers, with cytokines, especially Interleukin 6 (IL-6), being key. IL-6 is linked to conditions like cancer and sepsis, and its sweat levels correlate with blood levels. For certain conditions, threshold detection of IL-6 is crucial, as it simplifies the decision-making process by alerting users when the level exceeds critical points, signaling the need for clinical attention. This is especially relevant in applications such as sepsis diagnosis and monitoring, where IL-6 levels can indicate the progression of this life-threatening condition.
Among the various biosensing techniques for IL-6 detection, our work centers on electrochemical detection, for its high accuracy and sensitivity. In specific we focus on screen-printed three-electrode (SPCE) platforms, consisting of carbon-based working (WE) and counter electrodes and a silver/silver chloride reference electrode. Our low-cost, flexible, and in-house-made platform offers high accuracy and sensitivity for the detection of IL-6 using electrochemical readout techniques such as cyclic voltammetry (CV). While CV provides valuable information about electrochemical reactions, its ability to detect small changes in signal intensity is inadequate for accurate classification of physiological and pathological IL-6 levels in sweat (pg/ml scale). To overcome this challenge, sensitivity enhancement is necessary, which can be achieved by modifying the WE with gold nanoparticles (AuNPs) to improve electron transfer and increase the electrode's surface area, facilitating the immobilization of a higher density of aptamers. Another approach involves employing machine learning (ML) algorithms to further enhance detection accuracy by analyzing complex and process large datasets of electrochemical readings and distinguishing subtle differences in IL-6 concentrations that might not be evident through manual analysis of the CV readout.
In this work, we report the fabrication and characterization of a flexible SPCE system to detect IL-6 leveraging experimental CV and ML-based threshold detection. After screen-printing the SPCE system on a flexible polyethylene terephthalate substrate, AuNPs were electrodeposited on the WE and subsequently biofunctionalized with thiolated IL-6-sensitive aptamers to ensure selective detection. Then, we employed CV on different concentrations (5, 10, 20, and 200 pg/ml) of IL-6 in 1x phosphate-buffered saline (PBS). The cyclic voltammograms were processed by an ML algorithm for the threshold-based readout to detect physiological (<20 pg/ml) and pathological (>=20 pg/ml) IL-6 levels. The ML algorithm appears crucial as it can analyze complex CV datasets to adequately classify physiological and pathological IL-6 levels. This reasonably improves the reliability and efficiency of the detection process, surpassing the capabilities of conventional CV readout methods, such as oxidation peak current analysis.
In conclusion, our flexible and selective electrochemical platform, using AuNP-modified aptamer-functionalized SPCEs and applying CV without additional redox reporters combined with ML readout, competently classifies IL-6 concentrations into physiological and pathological ranges. This system shows great potential for real-time, non-invasive wearable diagnostics of pathological conditions, paving the way for advanced health monitoring solutions. Future expansion could include detecting a broader range of cytokines, each requiring specific calibration curves. Incorporating ML could further enhance the platform’s potential by optimizing calibration and identifying patterns across multiple cytokines, ultimately improving the ability to predict and understand critical pathological conditions.