Machine learning techniques have found widespread applications in bioinformatics. Such techniques meet the clinical requirements for various classification and prediction problems, which are difficult to model using traditional methods. For example, the current clinical practice uses wheelchair power seat function (PSF) to reduce seating interface pressure to prevent pressure ulcers. Typically, uniform clinic guidance is recommended to all patients. However, our preliminary study revealed that the response to PSF settings varied greatly from patient to patient with spinal cord injury (SCI). Therefore, it is highly desirable to develop an intelligent system that can predict the optimal PSF usage to reduce pressure ulcers risk for individual wheelchair users with SCI.
This is a cooperative research between the University of Central Oklahoma (UCO) and the University of Oklahoma Health Sciences Center (OUHSC). The goal is to develop an intelligent system that will achieve two specific aims: (1) To establish the relationship between wheelchair PSF usage and skin blood flow. Skin blood flow response to loading pressure has been regarded as an accurate way to determine the efficacy of seating conditions on reducing pressure ulcers risk. (2) To establish the relationship between PSF usage at various settings of duration and frequency and the reactive hyperemic response to sitting-induced loads. Specifically, we will use Artificial Neural Networks (ANNs) in Aim 1 to (a) return a set of favorable settings of PSF usage for an individual wheelchair user; (b) predict the best PSF usage that can significantly increase the skin blood flow for an individual wheelchair user; (c) cross-validate the outputs of two hypothesis functions to ensure the prediction quality. In addition, we will use ANNs in Aim 2 to predict the optimal duration and frequency to perform PSF for an individual wheelchair user. More importantly, this intelligent system will allow clinicians/staff to add clinical data and re-train the ANNs. As more data becomes available, the prediction results will become more and more precise. To our best knowledge, this will be the first intelligent system used to develop clinical guidance on PSF usage to reduce pressure ulcers risk.