Novel Analgesic Index for Postoperative Pain Assessment Based on a Photoplethysmographic Spectrogram and Convolutional Neural Network: Observational Study

Multimodal physiological sensing for the assessment of acute pain

Pain assessment is a challenging task encountered by clinicians. In clinical settings, patients' self-report is considered the gold standard in pain assessment. However, patients who are unable to self-report pain are at a higher risk of undiagnosed pain. In the present study, we explore the use of multiple sensing technologies to monitor physiological changes that can be used as a proxy for objective measurement of acute pain. Electrodermal activity (EDA), photoplethysmography (PPG), and respiration (RESP) signals were collected from 22 participants under two pain intensities (low and high) and on two different anatomical locations (forearm and hand). Three machine learning models were implemented, including support vector machines (SVM), decision trees (DT), and linear discriminant analysis (LDA) for the identification of pain. Various pain scenarios were investigated, identification of pain (no pain, pain), multiclass (no pain, low pain, high pain), and identification of pain location (forearm, hand). Reference classification results from individual sensors and from all sensors together were obtained. After feature selection, results showed that EDA was the most informative sensor in the three pain conditions, 93.2±8% in identification of pain, 68.9±10% in the multiclass problem, and 56.0±8% for the identification of pain location. These results identify EDA as the superior sensor in our experimental conditions. Future work is required to validate the obtained features to improve its feasibility in more realistic scenarios. Finally, this study proposes EDA as a candidate to design a tool that can assist clinicians in the assessment of acute pain of nonverbal patients.

A systematic review of neurophysiological sensing for the assessment of acute pain

Pain is a complex and personal experience that presents diverse measurement challenges. Different sensing technologies can be used as a surrogate measure of pain to overcome these challenges. The objective of this review is to summarise and synthesise the published literature to: (a) identify relevant non-invasive physiological sensing technologies that can be used for the assessment of human pain, (b) describe the analytical tools used in artificial intelligence (AI) to decode pain data collected from sensing technologies, and (c) describe the main implications in the application of these technologies. A literature search was conducted in July 2022 to query PubMed, Web of Sciences, and Scopus. Papers published between January 2013 and July 2022 are considered. Forty-eight studies are included in this literature review. Two main sensing technologies (neurological and physiological) are identified in the literature. The sensing technologies and their modality (unimodal or multimodal) are presented. The literature provided numerous examples of how different analytical tools in AI have been applied to decode pain. This review identifies different non-invasive sensing technologies, their analytical tools, and the implications for their use. There are significant opportunities to leverage multimodal sensing and deep learning to improve accuracy of pain monitoring systems. This review also identifies the need for analyses and datasets that explore the inclusion of neural and physiological information together. Finally, challenges and opportunities for designing better systems for pain assessment are also presented.

Photoplethysmography temporal marker-based machine learning classifier for anesthesia drug detection

Anesthesia drug overdose hazards and lack of gold standards in anesthesia monitoring lead to an urgent need for accurate anesthesia drug detection. To investigate the PPG waveform features affected by anesthesia drugs and develop a machine-learning classifier with high anesthesia drug sensitivity. This study used 64 anesthesia and non-anesthesia patient data (32 cases each), extracted from Queensland and MIMIC-II databases, respectively. The key waveform features (total area, rising time, width 75%, 50%, and 25%) were extracted from 16,310 signal recordings (5-s duration). Discriminant analysis, support vector machine (SVM), and K-nearest neighbor (KNN) were evaluated by splitting the dataset into halve training (11 patients, 8570 segments) and halve testing dataset (11 patients, 7740 segments). Significant differences exist between PPG waveform features of anesthesia and non-anesthesia groups (p < 0.05) except total area feature (p > 0.05). The KNN classifier achieved 91.7% (AUC = 0.95) anesthesia detection accuracy with the highest sensitivity (0.88) and specificity (0.90) as compared to other classifiers. Kohen’s kappa also shows almost perfect agreement (0.79) with the KNN classifier. The KNN classifier trained with significant PPG features has the potential to be used as a reliable, non-invasive, and low-cost method for the detection of anesthesia drugs for depth analysis during surgical operations and postoperative monitoring.

Deep convolutional neural network-based signal quality assessment for photoplethysmogram

Quality assessment of bio-signals is important to prevent clinical misdiagnosis. With the introduction of mobile and wearable health care, it is becoming increasingly important to distinguish available signals from noise. The goal of this study was to develop a signal quality assessment technology for photoplethysmogram (PPG) widely used in wearable healthcare. In this study, we developed and verified a deep neural network (DNN)-based signal quality assessment model using about 1.6 million 5-s segment length PPG big data of about 29 GB from the MIMIC III PPG waveform database. The DNN model was implemented through a 1D convolutional neural network (CNN). The number of CNN layers, number of fully connected nodes, dropout rate, batch size, and learning rate of the model were optimized through Bayesian optimization. As a result, 6 CNN layers, 1,546 fully connected layer nodes, 825 batch size, 0.2 dropout rate, and 0.002 learning rate were needed for an optimal model. Performance metrics of the result of classifying waveform quality into 'Good' and 'Bad', the accuracy, specificity, sensitivity, area under the receiver operating curve, and area under the precision-recall curve were 0.978, 0.948, 0.993, 0.985, 0.980, and 0.969, respectively. Additionally, in the case of simulated real-time application, it was confirmed that the proposed signal quality score tracked the decrease in pulse quality well.