A Relative Analysis of Machine Learning based approaches to Detect Human Pain Intensity using Facial Expressions

Pain is a distressing emotional and physical experience caused by tissue damage. Detection of pain is usually through self-reporting, and is not a feasible option for a patient in a coma or with impaired communication. Another technique called "observer’s report" is based on trained staff observing patients and detecting pain levels, but is subjective and error-prone. It is highly impractical to have hospital staff monitoring the patients around the clock. To solve the problem, automatic patient monitoring and alerting systems assist in continuous examination and detection of pain levels using facet reactions and improve the existing system. Pain is reliably indicated by facial expressions and is used in techniques for assessing pain by observation. The article analyzes, categorizes, and discusses methodologies in the literature produced. It provides specifics regarding the datasets, pre-processing techniques, Machine learning and Deep learning frameworks. It gives a comparative study of the same.