Abstract
To offer quality wellness care to patients, multi-parameter patient monitors (MPM) need a high accuracy for sensitivity, specificity, and overall classification. Nevertheless, it is likewise important to provide affordable healthcare by providing cheap MPMs using todays handheld computing and communication devices, and low complexity hardware. Support vector machine (SVM) is a vital classification process valuable for the improvement of MPMs for its high exactness and viability in foreseeing the status of patients. It is well known that non-linear kernel SVMs offer better performance, while the linear kernel SVM (LKSVM) are computationally very efficient. This makes the LKSVM particularly attractive for low cost implementations. In this paper, we demonstrate that mapping feature to a higher dimension using locality-constrained linear coding (LLC), added to that the framework by eluding the system reliant features using dimensionality reduction technique called principal component analysis (PCA) to make the framework durable, which improve the execution of MPMs using LKSVM. It was seen that the use of LLC-PCA has helped enhance the sensitivity by 3.27 % from the baseline system.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have