Anesthesia Level Prediction Using ML

Contribution of Anesthesia in modern medicine and surgery is critical, offering significant benefits to both patients and medical practitioners. According to a survey, the inaccuracy in predicting anesthesia levels has led to a 47 percent increase in the mortal rate among Clinical patients. This system proposes a solution to anesthesia management by integrating sensors and machine learning using Raspberry Pi. Machine learning algorithms such as Support Vector Machine (SVM), K-Nearest Neighbors (KNN), Artificial Neural Network (ANN), Gradient Boosting Machine (GBM) are used, and the best features that assist in anesthesia level prediction are selected from various feature selection methods. On evaluating the performances of various algorithms, KNN algorithm provides 97% accuracy, and the best features are found to be heart rate (BPM), blood oxygen saturation (SpO2), body temperature and blood pressure using feature selection methods. The proposed system is designed based on the KNN algorithm and it continuously monitors the vital signs acquired from the feature selection methods using MAX30100, DHT11, and PPG sensor. The prototype system designed decides the anesthesia level for the patient based on the vital parameters measured using sensor technology and the vitals and patients' history are displayed in a GUI. The proposed system signifies an advancement in automated anesthesia management, prominently projecting the potential of integrating ML models and sensor technology in effective implementation of medical procedures.