Beetle-inspired responsive photonic microgel assemblies for multi-sensing enhanced by machine learning.

Bioinspired photonic hydrogels hold promise as sensors; however, their use in triple-analyte sensing optical devices has been minimally explored. Temperature, serum Fe3+ levels, and X-ray doses are critical factors for predicting and monitoring medical and health issues. For the first time, poly(N-isopropylacrylamide-co-vinyl ferrocene) microgel-based assemblies, also known as interferometers, integrated with machine learning (ML) are introduced for detecting these analysts of interest. The interferometers demonstrate the exceptional capability for sensing with superior sensitivity, reversibility, fast responsivity, and simplicity. The collected optical signals are utilized by various machine learning algorithms, including k-nearest neighbor (KNN), decision tree (DT), support vector machine (SVM), and feedforward artificial neural network (FANN). The FANN model exhibits superior ML capabilities compared to others, achieving an accuracy of 95.24 % in predicting temperature and 100 % for both Fe3+ concentrations and X-ray doses. As proof of concept, the interferometers detect the temperature of water drops, Fe3+ concentration in serum, and clinical radiotherapy levels of X-ray doses. The results are consistent with those from commercial infrared thermography, clinic ferrozine colorimetric methods, and radiotherapy treatment planning systems. This work demonstrates the advantages of using this interferometer in practical applications via ML algorithms.