Abstract

Pain assessment is critical for gaining valuable insights into a patient's health status and predicting recovery outcomes. The subjective nature of pain and the influence of individual, psychological, and social factors make assessment difficult. Pain assessment is primarily based on self-reporting or expert observation; however, both methods have inherent limitations. Self-reporting may lack reliability and feasibility for specific patients. In contrast, expert observation is inherently subjective and requires experienced personnel, making it impractical in the context of rising inpatient numbers and overburdened healthcare providers. As a result, This research study proposes “HAIEN,” an intelligence system designed to autonomously detect and categorize pain levels in inpatients using facial expression analysis. The “HAIEN” application aims to provide a valid and reliable pain assessment, allowing healthcare providers to make informed treatment decisions and ensure ongoing patient care. Two classifier models, kNN and SVM, were trained on the UNBC-McMaster Shoulder Pain Database. The two classifiers used three feature extraction methods: VGG16, EfficientNetB3, and InceptionV3. The findings show that these models successfully captured facial movements and correctly identified pain. Using Artificial Intelligence technology in the “HAIEN” application improves the pain assessment process and patient health outcomes.

Key words: Pain assessment; Facial expression; Deep learning; Data analysis.