Oct 5 2020
A new study presented at the ANESTHESIOLOGY® 2020 annual meeting suggests that artificial intelligence (AI) utilized in machine learning models can predict the patients who are at the highest risk of severe pain following surgery and assists in identifying who would profit most from personalized pain management plans that involve using of non-opioid alternatives.
Image Credit: Ryzhi/Shutterstock
Certain patients experience more severe pain following surgery and require higher doses of opioids for longer periods, which raises their risk of opioid abuse disorder.
By getting to know which patients are at a greater risk of severe post-surgical pain, physician anesthesiologists can make an anesthesia plan by making use of non-opioid alternatives like epidurals, nerve blocks, and other medications to more efficiently address pain and decrease the requirement for opioids.
At present, physicians make use of laborious questionnaires to find patients at greater risk of severe post-surgical pain, inquiring about their history of depression, sleep quality, and anxiety.
In this research, the team wanted a quicker and more effective technique that involves the use of machine learning, where a system learns and transforms based on data it is offered. They made three machine learning models that examined the electronic medical records patients, which found that female gender, younger age, pre-existing pain, higher body mass index, and prior usage of opioids were the most predictive factors of post-surgical pain.
We plan to integrate the models with our electronic medical records to provide a prediction of post-surgical pain for each patient. If the patient is determined to be at high risk for severe post-surgical pain, the physician anesthesiologist can then adjust the patient’s anesthesia plan to maximize non-opioid pain management strategies that would reduce the need for opioids after surgery.
Mieke A. Soens, MD, Study Lead Author and Anesthesiologist, Brigham and Women’s Hospital
Soens is also an anesthesiology instructor at Harvard Medical School, Boston.
As part of the two-part study, the team analyzed data from 5,944 patients who had undergone an extensive range of surgeries, such as prostate surgery, hip replacement, hysterectomy, and gallbladder removal.
Among them, 1,287 (22%) had consumed 90 morphine milligram equivalent (MME) in the first 24 hours following surgery, which is regarded as a high dose. In the study’s first part, they utilized 163 possible factors to anticipate high pain post-surgery, depending on a literature search and consultation with experts.
From there, they made three machine learning algorithm models (artificial neural networks, random forest, and logistical regression) that mined the medical records of the patients and reduced the 163 predictor factors down to those which most precisely anticipated patients’ pain severity and possible opioid requirements following surgery.
In the second part, the researchers compared what was predicted by the models with actual opioid use in the same patients. It was found that overall, all three models had similar predictive accuracy: 80% for artificial neural networks and 81% for logistical regression as well as random forest techniques.
This implies that the models precisely determined which people had a greater likelihood to have severe pain and required higher doses of opioids around 80% of the time.
Electronic medical records are a valuable and underused source of patient data and can be employed effectively to enhance patients’ lives. Selectively identifying patients who typically need high doses of opioids after surgery is important to help reduce opioid misuse.
Mieke A. Soens, MD, Study Lead Author and Anesthesiologist, Brigham and Women’s Hospital