The management of a patient’s subconscious pain response, known as “nociception,” during surgery can significantly impact the intensity of post-operative side effects and the need for further pain management. However, measuring pain is inherently subjective—especially when patients are unconscious.
In a groundbreaking study, researchers from MIT and Massachusetts General Hospital (MGH) have developed a set of statistical models to objectively quantify nociception during surgery. The findings, published in the Proceedings of the National Academy of Sciences, aim to assist anesthesiologists in optimizing drug dosages and minimizing post-operative pain and side effects.
The research is based on data collected from 101 abdominal surgeries at MGH, covering 18,582 minutes of surgery. Led by Sandya Subramanian, now a professor at UC Berkeley and UC San Francisco, the team analyzed data from five physiological sensors and recorded 49,878 distinct nociceptive stimuli, such as incisions or cauterization. They also tracked the drugs administered, including dosage and timing, to account for the effects on pain and cardiovascular responses.
The ultimate goal is to provide anesthesiologists with real-time, objective data to guide their decisions on pain-control drugs. Overdosing these drugs can lead to side effects like nausea and delirium, while underdosing can leave patients in significant post-operative pain.
“Sandya’s work has helped us establish a principled way to understand and measure nociception (unconscious pain) during general anesthesia,” said senior author Emery N. Brown, an anesthesiologist at MGH and professor at MIT and Harvard Medical School.
Subramanian’s research began as a doctoral project at MIT in 2017. Unlike prior attempts, which relied on limited measurements such as electrocardiograms (ECG) or lab-based experiments, this study collected multi-sensor data during real surgeries. The team tracked heart rate, skin conductance, skin temperature, and respiration to build more accurate models for pain detection, while factoring in the effects of anesthesia.
Heart rate changes and electrodermal activity (EDA) were key indicators, as they are linked to the body’s “fight or flight” response to pain. However, some drugs used in surgery can affect heart rate, so the team used additional measures like PPG (an optical heart rate sensor) and skin temperature to ensure accurate detection.
Through rigorous statistical analysis, the researchers developed models that predict nociception by analyzing the combined indices from all physiological signals.
The team developed five versions of the model, with some incorporating supervised learning—where the model was fed information about specific nociceptive events—and others using an unsupervised approach. The supervised models that incorporated drug information performed best, with the random forest statistical approach emerging as the most effective.
A key achievement was the performance of the unsupervised model, which still outperformed existing industry standards like the ANI model (which tracks ECG). This suggests that there is an objectively detectable signature of nociception, even across different patients and without prior information about painful events.
“This is an important step toward defining a metric to track nociception without including nociceptive ‘ground truth’ information,” the authors wrote, emphasizing the model’s potential for scalable, real-time clinical applications.
The next phase of research will focus on refining the models for real-time use during surgeries. The long-term vision includes closed-loop systems, where the models could automatically adjust anesthesia and pain-control drug dosages under the supervision of anesthesiologists.
The team believes their work could eventually lead to better pain management not only in surgeries but also in other clinical settings like intensive care units (ICU). The models could pave the way for objective, data-driven approaches to pain management, improving patient outcomes while reducing the reliance on subjective assessments.
“Our study is an important first step toward developing objective markers to track surgical nociception,” the authors concluded, pointing to a future where surgical pain can be managed with greater precision and fewer side effects.
By Impact Lab
