Model-based and data-driven with multiscale sum of squares double EWMA control chart for fault detection in biological systems

The objectives of this paper will be sought. First, an enhanced technique that can accurately model biological processes will be developed. To deal with scenarios where a process model is available, the particle filter method will be developed to better handle the nonlinear and high-dimensional state estimation problem. Second, a multiscale sum of squares double exponentially weighted moving average (MS-SS-DEWMA) chart will be applied to the monitored residuals in order to enhance the fault detection abilities. The advantage of MS-SS-DEWMA chart is twofold: (1) The SS-DEWMA chart uses the sum of squares statistics; it simultaneously monitors the process mean and variance in a single chart. It has presented better performance than the classical EWMA-based charts. (2) The multiscale data representation can be used as an effective tool for reducing noise from a signal's time series. The effectiveness of the proposed strategy is validated using a synthetic and simulated Cad system in Escherichia coli (CSEC) data. When the simulated CSEC model is used, the developed approach is applied for monitoring some of the key variables involved in the CSEC model. The proposed strategy is also applied to detect diseases using genomic copy number data through better detection of aberrations in the genetic information of patients, which can help medical doctors make more accurate diagnosis of diseases.