Enhancing Pain Level Assessment in Post-Surgery Patients Using Artificial Intelligence Algorithms

Control performance decreases significantly in the presence of uncertainty in variable availability, measurement noise, or instrumentation failure. In cluttered environments such as the Post Anesthesia Care Unit (PACU), clinical measures are often influenced by noise and artifacts. An important component in post-operative treatment is the assessment and management of pain levels. However, reliable information is critical for clinically relevant results and improved patient outcomes. From a control engineering point of view, variables are often estimated and interpolated to allow a suitable flow of feedback information to control loops for the optimization of drug dosages. In this context, Artificial Intelligence (AI) stands as a promising tool to augment pain level assessment. This study introduces and compares two AI-based approaches for predicting continuous Numerical Rating Scales (NRS) based on heart rate (HR) data. The first approach uses polynomial regression, lasso regression, and ridge regression, while the second employs an LSTM model. Notably, the AI prediction model, independent of traditional interpolation techniques, outperforms the approach relying on interpolation. The proposed AI-based method holds promise for continuous estimation and can serve as an estimator for model-based control. This proof of concept study underscores the potential of AI tools to enhance pain level assessment.