A Novel Rapid Bacterial Infection Screening Multisensor System With Feature Selection and Sensor Array Optimization

In this paper, a novel multi-sensor detection system framework for the rapid screening of bacterial infection is proposed. To capture the dynamic information of sensor response curves, eight features, such as time-domain and frequency-domain features, are extracted for each sensor. Additionally, a novel feature selection algorithm based on adaptive similarity and latent semantics (ASLSFS) is employed to eliminate irrelevant features in the initial feature set. Due to the redundant information and noise introduced by the sensors' broad-spectrum response characteristics and hardware circuit interference, a dynamic information change weighting array optimization algorithm (DICWAO) is developed, which leverages the impact of adding candidate sensor features on the shared information among previously selected sensor features, candidate sensor features, and class label. The experimental results validate the effectiveness of the designed system. Comparative analysis with existing algorithms verifies the effectiveness of the developed feature selection algorithm and array optimization framework.