A portable nanobiosensor with machine learning: enabling multi-cytokines point-of-care testing and early identification of sepsis immunoparalysis

Abstract Sepsis immunoparalysis (SIs) is a major factor contributing to organ dysfunction. However, emergency departments (EDs) and intensive care units (ICUs) currently lack a rapid workflow for early identification of SIs. Herein, an innovative strategy for early SIs identification is proposed. We construct a portable nanobiosensor based on the Nanozyme-Linked Immunosorbent Assay (NLISA) for ultrasensitive and specific Multi-Cytokines quantification, benefiting from its advantages of parallel, independent, rapid, and portable detection. Machine learning is applied to process Multi-Cytokines concentration data measured under varied conditions and build an early identification model for rapid and effective sepsis immunoparalysis diagnosis. This method achieves point-of-care testing (POCT) with small plasma sample volumes (30 µL), with a detection limit as low as 3.8 fg mL − 1 . Notably, machine learning models constructed from different cytokine combinations can achieve high diagnostic performance, accurately distinguishing sepsis immunoparalysis, sepsis non-immunoparalysis, and healthy donors, while addressing diverse clinical needs and cost-effectiveness considerations, enabling personalized diagnostic strategies tailored to distinct patient populations and resource-constrained settings. This opens up the potential application of the nanobiosensor with machine learning for POCT of Multi-Cytokines detection and sepsis immunoparalysis early identification.