Machine learning intelligently selects feature values to construct a sensor array based on tri-functional Mn-doped...

Herein, a novel tri-functional Mn-doped covalent organic polymer nanozyme (Mn-COP) with peroxidase-like, oxidase-like, and laccase-like activities was synthesized. The strong reducing power of flavonoids inhibited the tri-enzymes-like activity from Mn-COP, with the trend increasing as a function of reaction time. The Mn-COP endowed the sensor array with improved accuracy due to three different chemical reactions used as recognition channels. To assess the significance of reaction time in optimizing eigenvalues, the Random Forest algorithm (RF) was employed. The nanozyme sensor array effectively distinguished various flavonoids. The incorporation of the RF enabled accurate identification and prediction of 7 flavonoids within the concentration range of 10-500 μM present in a variety of traditional Chinese medicines. Overall, a key innovations of this work is the introduction of a more scientific and intelligent approach to selecting feature values, thereby opening a promising avenue for the development of sophisticated nanozyme sensor arrays.