Utilizing autonomic features with machine learning techniques can
provide an effective alternative for periodontal pain detection.
The fusion of autonomic indices using machine learning techniques serves as a promising detection method to identify acute periodontal pain, as per research published in the Journal, Physiological Measurement.
Acute periodontal pain is a problem associated with the periodontium or related structures. Approaches to managing pain are the primary concern in dental care. The study helps to elucidate the machine learning strategies that can detect the brief periodontal pain easily in means of recordable physiological signals among 47 patients. These patients went through periodontal probing and showed each case of pain perception using a push button. Concurrently, physiological signals were noted and, afterwards, computation of autonomic indices were done.
A pain indicator based on the fusion of the numerous autonomic mechanisms was obtained using the autonomic indices as input characteristics of a classifier. The selection of seven patients for the test set was done randomly. The remaining data were used for the several classifiers and feature combinations verification using leave-one-out-cross-validation. The random forest classifier during the validation process, using frequency spectral bins of the ECG, PPG amplitude, SPI as features, and wavelet level energies of the ECG and PPG emerged as the best pain detection algorithm. The independent test dataset of the algorithm's final test produced 70% and 71% of specificity and sensitivity respectively.
As per these
findings, the fusion of autonomic indices using machine learning methods is an
excellent choice for brief pain perception detection.
Physiological Measurement
Detection of acute periodontal pain from physiological signals.
Daniel Teichmann et al.
Comments (0)