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

Abstract

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. Graphical abstract

Effect of intracranial pressure on photoplethysmographic waveform in different cerebral perfusion territories: A computational study

“…Considering the individual difference in waveform which may involve other confounders, we did not include the analysis of focal waveform features, e.g., the location of maximal/minimal first or second derivatives. These features may reflect important physiological information including neural activities ( Khalid et al, 2022 ) and cardiovascular pathophysiological changes ( Elgendi et al, 2018 ). Second, the Windkessel models were highly simplified where the local hemodynamic changes within an arterial segment or a perfusion territory could not be reflected.…”

Implantable Intracranial Pressure Sensor with Continuous Bluetooth Transmission via Mobile Application

“…Our presented device offers a few distinct benefits, such as the ability for patients to measure ICP remotely and provides continuous monitoring on a mobile application without a wearable component, as it is integrated into their existing shunt catheter. Other noninvasive ICP monitors, such as the device by Abay et al, differ as they require a wearable component that may be sensitive to motion artifact contact pressure and require full time application of the wearable component by the patient to obtain continuous data [ 22 ]. Additionally, our presented device offers a specific benefit of integration of the pressure sensor into the lining of the proximal catheter, targeting a specific population of shunted hydrocephalus patients and aiding in the workup of shunt failure.…”

Development and validation of a combinatorial MRI sequence-based radiomics model for preoperative prediction of microsatellite instability status in rectal cancer

“…The KNN algorithm is simple and easy to apply, with few assumptions regarding data distribution, and it is suitable for various types of data, including medical imaging data. Compared with other machine learning algorithms, such as the SVM and decision tree, over tting is less likely to occur with the KNN algorithm [22] .…”