MoS2-Based Multifunctional Sensor for Both Chemical and Physical Stimuli and Their Classification Using Machine Learning Algorithms

This report demonstrates for the first time the direct growth of the molybdenum disulfide (MoS
₂
) on flexible pencil eraser substrate by cost-effective hydrothermal synthesis procedure and its application as a multifunctional sensor to monitor human appendage movements and breath pattern for smart personal health care services. The gauge factor calculated was 7.91, which is comparable to the strain sensors fabricated with sophisticated techniques and relatively better when the cost is a major constraint. One of the important roadblocks towards the development of multifunctional sensor is the inability of the frontend to identify stimulus. To overcome this, the categorization of sensor output data for a reliable front-end processing of the sensor data was performed using various machine learning (ML) algorithms such as Logistic regression, naive Bayes, k-nearest neighbors, support vector machine, decision tree, and random forest. The performance evaluation of the algorithms was tested using 10-fold cross validation (CV) method. The decision tree algorithm was able to classify both breath and strain sensory data with 95.57% accuracy, followed by random forest algorithm with 94.77% without tuning parameters. Decision tree could classify with 95.96% accuracy using 10-fold CV without tuning parameters whereas k-nearest neighbors gives 96.63% with 10-fold CV and tuning parameters. Further, the fabricated sensor was integrated on different parts of human body and corresponding movement were monitored. After 500 bending cycle studies, negligible change in the sensor performance suggesting the excellent durability of the device.