AI-powered pain assessment: Korean researchers develop new surgical tool

The rapid advances of artificial intelligence (AI) present valuable opportunities for better patient outcomes and optimized surgical care. Pain assessment in surgery used to be a subjective matter, depending on patient reports and the opinions of doctors (and other experts). However, recently, researchers at Asan Medical Center (AMC), Seoul, South Korea, have successfully developed a system driven by AI to measure pain objectively in patients during surgery and recovery. This technology will exert major impacts on healthcare in terms of pain alleviation for patients (especially for those who are unconscious or unable to communicate) and post-surgery patient care in general.

As AI continues transforming healthcare, innovations like the AI-powered pain assessment model developed by AMC highlight the growing intersections between technology and patient care. For aspiring nurses, to have an understanding of these advancements is essential, as many of the best online ABSN programs now incorporate AI applications in the healthcare context, providing students with the necessary knowledge and skills to prepare for a future where technology will play a vital role in patient assessment and management.

While AMC’s system represents a major leap forward in the use of AI in surgical care, it was not the first case of the use of machine learning in pain assessment.

For example, several studies have used AI to analyze facial expressions for pain assessment. These systems were found to automatically detect pain successfully with relatively high accuracy in more than 95% of subjects. Other studies have used AI approaches to analyze clinical notes and patient records containing pain assessment information to identify all components relating to pain classifications and severity. Further applications of AI have been for patients with severe dementia and those who cannot verbalize or communicate, where pain assessment is given through facial recognition, smart computing, etc.

Notwithstanding this, the system devised at AMC involves tracking patients’ heart rate, blood pressure, and blood volume changes during surgery, where the machine learning algorithm is utilized to analyze these measurements. AMC’s study involved 242 surgery patients, with six variables relating to pain prediction being selected and inputted into the system as a way to confirm pain occurrence both during and post-surgery. Researchers found that the AI-based model matches the accuracy of existing models for intraoperative pain assessment (pain experienced during a surgical procedure when the patient is under either general or local anesthesia) at a level of 83%. However, it greatly outperformed in postoperative pain assessment (pain expected after surgery) at a level of 93% accuracy, while existing models only had 58% accuracy.

Furthermore, throughout the study, two more predictors – systolic upper limit variability (changes in the highest blood pressure reading (systolic number)) and pulse width (how long blood pulse takes to move through arteries in each heartbeat) – were found to be of greater significance than what existing assessment models had detected, which can be crucial for the development of more effective postoperative pain management strategies.

Healthcare providers have relied on traditional pain scales such as the numeric rating scale (NRS), which is an 11-point numeric scale ranging from ‘0’ (no pain) to ‘10’ (extreme pain), or the visual analog scale (VAS), which is a linear measure doctors use to record pain progression, to assess a patient’s pain levels.

Of course, these tools have been extremely useful due to the obvious fact that they have persisted for so long, but with the continual advancements of health along with the complexification of illnesses, diseases, or conditions, their limitations are becoming more obvious. For example, these measurement methods can be difficult to use properly if the patient has cognitive impairments or communication difficulties, leading to inaccurate pain reports. Contrarily, this is where AI technologies such as those developed by AMC can have an opportunity to shine and make a difference.

According to Dr. Byong Moon Choi, professor at the AMC Department of Anesthesiology, the machine learning technology can enable doctors to ‘objectively evaluate the level of pain in unconscious patients, such as those under sedation or those who have undergone endotracheal intubation’, as well as become an important tool for ‘future personalized pain management.’ Using facial and body language recognition or other physiological cues to estimate pain levels can pave the way for more objective and reliable pain scores, especially for demographics that cannot self-report pain effectively. Such algorithms have been directly trained on huge datasets of pain-related behaviors, which also means they can detect subtle nuances or complexities in patients that human observers cannot.

Traditional pain assessment tools can also be influenced by racial and cultural biases, which can potentially result in poor pain management and worse health outcomes; using AI can mitigate these factors and allow for more targeted and responsive pain management strategies. One major thing that scores AI-driven assessments is personalized pain management strategies, which can potentially reduce dependency on medicines such as opioids. But of course, these systems are still in their early stages of development, and further validation is still needed before officially implementing them as a tool for practice.

AMC is not the only enterprise researching into the uses of AI technology in pain assessment and management – in fact, institutions around the world are looking into ways to cover the area more smartly. For example, PainChek is an Australian AI-based company that has introduced a mobile application to assess pain levels through facial recognition for elderly and paediatric patients; NEC Corporation in Japan has offered the use of AI to assist self-care in spotting areas of chronic lower back pain; and AppliedVR, a U.S.-based startup, has made a virtual-reality system that can manage chronic pain.

So what does this mean for the healthcare sector with all these quickly emerging technologies? They will disrupt the whole field of pain management, but they also represent opportunities for hospitals and clinics to allow doctors and nurses to make decisions faster and more accurately, and rely on the better features of real-time pain control, personalized pain management, as well as improved precision in patient care.

Of course, there will be resistance from some healthcare institutions, but it will be continually embraced further on when more and more trials are conducted and the results they yield become clearer. AI should be viewed as a tool that enhances the skills of doctors, rather than a doomsday scenario where AI will end medicine as we know it. When these technologies are used correctly and effectively, it is a win-win situation for both the doctor and the patient, as it means less pain, quicker recoveries and a better quality of life overall.