Intelligent algorithm based on deep learning to predict the dosage for anesthesia: A study on prediction of drug...

Area of Science:

• Anesthesiology
• Artificial Intelligence
• Medical Informatics

Background:

• Accurate anesthetic drug dosage is critical for surgical safety.
• Individual patient variability and complex physiology challenge precise dose prediction.
• Deep learning offers a promising approach to enhance anesthetic management.

Purpose of the Study:

• To develop and validate a deep learning model for predicting anesthetic drug dosage.
• To improve the accuracy of both preoperative and intraoperative anesthetic dose estimation.
• To provide anesthesiologists with a tool for better clinical judgment and drug control.

Main Methods:

• An artificial neural network with SELU activation and weighted regularization for preoperative dose prediction.
• A Convolutional Neural Network (CNN) with a 7x1 kernel for intraoperative feature extraction.
• A Long Short-Term Memory (LSTM) network, enhanced with bidirectional propagation and an attention module, for intraoperative dose prediction.

Main Results:

• The proposed deep learning model significantly reduced Mean Absolute Percentage Error (MAPE) for preoperative (15.83%) and intraoperative (12.25%) dose predictions compared to traditional methods.
• Achieved improved prediction accuracy, with values of 0.887 for preoperative and 0.915 for intraoperative doses.
• Demonstrated the model's effectiveness in handling unbalanced sample data and extracting relevant patient features.

Conclusions:

• The developed intelligent anesthesia prediction algorithm effectively forecasts patient anesthetic drug requirements.
• The model assists anesthesiologists in making informed clinical decisions regarding drug dosage.
• Implementation of this algorithm can contribute to safer and smoother surgical procedures.