AUTOMATIC PAIN RECOGNITION SYSTEM FOR DENTAL PATIENTS USING MACHINE LEARNING

Abstract

Dental patients often struggle to effectively communicate their pain during treatment where they relies on their gestures and physical movements. This may disturb the treatment process from the dentists. This patient self-reporting method also can be subjective and inconsistent. In this study, we present a machine learning-based automatic pain recognition system designed to objectively recognize the pain levels in dental patients. The bio-signals comprising heart rate obtained from Electrocardiography (ECG), muscle activities extracted from Electrocardiography (EMG), and brain activity derived from Electroencephalography (EEG) have been extracted using the AD-8232 sensor for ECG and the BITalino sensor kit for EMG and EEG recordings. These signals have been normalized and classified by a machine learning classifier into "High pain," "Mild pain," and "No pain" categories. The system underwent training and testing using the bio-signals as input data and the pain levels as the output. Python served as the programming language for machine learning training, while the open-source integrated development environment (IDE) Jupyter Notebook has been employed as the primary platform for the model development. Eight distinct machine learning algorithms have been utilized for model training, including Random Forest Classifier, K-Neighbours Classifier, Bagging Classifier, Decision Tree Classifier, Logistic Regression, SGD Classifier, Linear SVC, and ADA Boost Classifier. Random Forest model demonstrated the best performance, achieving the highest accuracy of 65.052%. This research contributes to the development of automatic pain assessment in dental treatment.