Machine learning-assisted BMOFs-derived 1D NiCo2O4 nanozyme photoelectrochemical detection of RBP4 for type 2 diabetes diagnosis

Abstract

Retinol binding protein 4 (RBP4) is a potential marker for the diagnosis of Type 2 diabetes mellitus (T2DM). Due to the low sensitivity of RBP4 detection technology, the application of RBP4 in the early diagnosis of T2DM remains to be further explored. Herein, a highly sensitive photoelectrochemical (PEC) biosensor based on chitosan-modified covalent organic framework (COFC) and bimetallic metal organic frameworks-derived one-dimensional (1D) NiCo₂O₄ nanorods is developed for the detection of serum RBP4, using machine learning to assist in intelligent diagnosis of T2DM. Benefiting from the excellent photoelectric properties of COFC and the multifunctional signal amplification of 1D NiCo₂O₄ nanozyme, this PEC biosensor has a broad detection range from 10 to 10⁷ fg/mL and a low detection limit of 8 fg/mL. Importantly, the established PEC platform for RBP4 analysis in human serum can effectively distinguish healthy individuals from T2DM patients. Moreover, this machine learning is employed to probe the hidden potential pattern in the proposed PEC technology data, and the accuracy of machine learning for T2DM intelligent diagnosis can reach 95.8 %. The combination of PEC biosensor and machine learning provides an efficient tool for RBP4 analysis, aiding in the early diagnosis of T2DM.