SiNW-ISFET Sensor Modeling Using the k-Nearest Neighbor Machine Learning

The silicon nanowire (SiNW)-based ISFET sensor is experiencing tremendous development due to its multiple attractive advantages (small size, low cost, robust, and portable real time). The ISFET sensor is widely used for pH detection but also for patient health monitoring, cancer detection, water and soil content analysis, and other applications. Currently, few predictive models are used to estimate the performance of the ISFET sensor and determine the most influential parameters. In this study, a new model based on the machine learning (ML) technique named k-nearest neighbor (KNN) is developed to estimate the performance of silicon nanowire ISFET. The KNN model was tested for different values of training data and compared with experimental results. A good agreement was observed with the experimental measurements. Moreover, the test results show that the accuracy percentage of the KNN model with 90% training data could reach up to 99.96% and could meet the practical demand.