A system that measures a patient’s degree of pain by examining brain activity from a portable neuroimaging device has been created by researchers. The method may assist medical professionals in identifying and managing pain in unconscious and nonverbal patients, potentially lowering the risk of postoperative chronic pain.
The researchers outline a technique to measure patient discomfort in a paper. They achieve this using a newly developed neuroimaging method called functional near-infrared spectroscopy (fNIRS). Sensors placed all over the cranium measure oxygenated hemoglobin concentrations that signify neuron activity.
The prefrontal cortex, which significantly impacts how pain is processed, can be measured by the researchers using a small number of fNIRS sensors placed on a patient’s brow. The researchers created individualized machine-learning models using the measured brain signals to find patterns in the amounts of oxygenated hemoglobin connected to pain reactions. When the sensors are in position, the neuroimaging device models have an accuracy of about 87 percent when determining whether a patient is in pain.
A crucial aspect of the model is that it automatically creates “personalized” submodels that take high-resolution features from specific patient subpopulations. On average answers from the total patient population, one model in machine learning learns classifications — “pain” or “no pain” — based on these responses. But with diverse patient populations, that generalized strategy can decrease accuracy.
Related Content: Exceptional Points For Sensitive Microring Sensors