Pain is an unpleasant feeling that can be sharp or dull. There are often physical reasons for severe pain. However, emotional, psychological, and social factors can affect our perception and response to pain. “Pain is usually made up of a physical and a psychosocial component,” says Noemi Gozzi, ETH Zurich doctorate student (1^✔ ✔Trusted Source
Unraveling the physiological and psychosocial signatures of pain by machine learning

Go to source

).
Doctors usually consider this while recommending a course of treatment. However, it hasn’t been easy to differentiate one component from the other. Based on the patient’s complaints, doctors often use simple procedures to assess pain and its severity.

This frequently results in unrelated treatments. Opioid medications continue to see widespread use despite their drawbacks, side effects, decline in effectiveness, and the possibility of developing an addiction to the drug or overdose death.

Personalizing Treatment Approaches

In recent years, Stanisa Raspopovic’s group at ETH Zurich, of which Gozzi is a member, has worked with researchers at Balgrist University Hospital in Zurich to develop an approach that can clearly distinguish and quantify physical and psychosocial components of pain. They have published their new method in the current issue of the journal Med. Raspopovic was a Professor of Neuroengineering at ETH Zurich until recently.

“Our new approach should help physicians to assess patients’ pain more individually and thus offer them more tailored personalized treatment in the future,” Raspopovic says. If the pain is primarily physical, doctors are likely to focus their treatment on the physical level, including the use of medications or physiotherapy. If, on the other hand, psychosocial factors play a major role in the patient’s experience of pain, it may be indicated to positively change the perception of pain with psychological or psychotherapeutic support.

Analyzing Pain Perception

To develop the new method, the researchers analyzed data from 118 volunteers – including people with chronic pain and healthy controls. The researchers asked the study participants in detail about their perception of pain and any psychosocial characteristics such as depression, anxiety, and fatigue and how often they were in so much pain that they were unable to go to work.

In addition, the researchers recorded how well the participants can distract themselves from pain, and the extent to which pain gets them brooding or makes them helpless and causes them to overestimate the pain.

The researchers used standardized measurements of sensations of spontaneous pain to compare the subjects’ perceptions of pain. Participants were administered small, non-dangerous but painful pulses of heat on their skin. To record the physical reaction to the pain, the researchers measured the study participants’ brain activity using an electroencephalogram (EEG) and the electrical conductivity of the skin.

The latter changes depending on how much someone is sweating and it is used to measure stress, pain, and emotional arousal. Finally, the extensive dataset included the diagnoses of the study participants, which were made by the researchers at Balgrist University Hospital.

Precision Medicine Is Made Possible by Machine Learning

Machine learning helped the researchers to analyze a large amount of data, clearly distinguish between the two pain components, and develop a new index for each. The index for the physical component of pain indicates the extent to which the pain is caused by physical processes.

The index for the psychosocial component indicates how strongly emotional and psychological factors intensify the pain. Finally, the scientists validated these two factors using the participants’ comprehensive measurement data.

The new method, with its combination of measuring body signals, self-disclosure, computerized evaluation, and the resulting two indices, is intended to help physicians treat pain. “Our method enables physicians to precisely characterize the pain a particular person is suffering so they can better decide what kind of targeted treatment is needed,” Gozzi says.

The researchers at ETH Zurich and Balgrist University Hospital are continuing this project; together with the Clinique romande de réadaptation in Sion and the spinal cord injury department of a hospital in Pietra Ligure, Italy: they’re investigating the clinical relevance of the new method in a long-term study.

Reference:

1. Unraveling the physiological and psychosocial signatures of pain by machine learning – (https://www.cell.com/med/fulltext/S2666-6340(24)00298-8?)

Source-Eurekalert