Reading a mind in pain

Share this post:

Editor’s note: This article is by Dr. Guillermo Cecchi of IBM Research’s Computational Biology Group. 
Pain, whether dealing with a healthy or sick person, is an enormous part of medicine. Yet it is poorly understood. Consider this: we are often asked to rate our perception of pain intensity, by reporting it in a scale between 1 and 10. In this way, pain can be measurable. However, pain is a highly subjective phenomenon, heavily determined by perceptual and cognitive mechanisms, so that individuals’ pain perception levels vary widely.
Chronic pain affects at least 10 percent of the population
[source: Harstall C and Ospina M (June 2003). “How Prevalent Is Chronic Pain?”. Pain Clinical Updates, International Association for the Study of Pain XI (2): 1–4.]

But there are patterns. I have been working with with my colleague Irina Rish, and Northwestern University’s Dr. Apkar Apkarian at his Pain and Emotion Lab for several years to find those patterns in function Magnetic Resonance Imaging (fMRI) scans. And this month, our paper Predictive Dynamics of Human Pain Perception detailed findings based on experiments carried out to understand the emergent properties of functional brain networks shown on these scans.
We demonstrated that subjective responses to pain can be captured by a unique model, through fMRI, where individual differences are determined by a handful of parameters. A physician could then use this knowledge to personalize patient diagnosis and treatment.
Measuring pain
To be clear, none of the study’s 26 participants were harmed. We measured brain activity based on stimuli delivered via a thermal plate at different points on the skin. They were told that the plate would change temperature, but not when or to what temperature.
We describe this in the paper in more details, but the mind displays three features when in pain:
  • “First, the pain must signal the threat of tissue damage. This is determined by the current value of the skin temperature. The signal magnitude must consistently increase with the temperature, although not necessarily linearly (as in fact, tissue damage is not linear with temperature).
  • “Secondly, this signal magnitude [registered in the brain] must anticipate the possibility of damage – sounding the alarm of an imminent threat, given the recent history of temperature values, independently of the current temperature. This information can also be partially captured by the skin temperature’s rate of change.
  • “Finally, given its powerful hold on behavior, the intensity of pain perception must quickly decay once the threat of damage disappears, so as not to interfere with [a person’s] other ongoing mental states.”
The raw value of the temperature is just one of the drivers of perception. Our work also showed that there are other components, such as the need to anticipate potential damage, even when the current temperature is comfortable, and the need to forget quickly, even when the current temperature is high, but falling.
Our work represents the first model of pain as a perception with a predictable, deterministic signature. Beyond pain, there are very few examples of similar models for other mental processes, and our work s
hows the power of simultaneously modeling mental states and brain activity.
And the brain activity of our participants reflected these components.
In fact, the brain areas that can be used to infer the raw value of the temperature are different from those that are used to infer pain perception. Moreover, combining the inferred pain perception from fMRI readings with a prediction based on the inferred temperature gives us the best predictive model.
Helping doctors help your pain
By showing that subjective responses to pain can be constrained by a handful of parameters, doctors can personalize patient diagnosis and treatment. Our “mind reading” approach of using fMRI also implies that it is possible to infer the level of pain a person is suffering even when he or she cannot report it verbally, or otherwise. We hope that it will also be the basis of a more accurate prognosis for those patients with risk of developing chronic pain (for instance, those with prolonged back pain).
The next steps in this study are to find out how the model changes for patients with chronic pain, in the hope that it will reveal specific mechanisms disrupted by the condition. And we will also study pharmacological effects on the model to answer questions such as: which part of the model is affected by analgesics and anesthesia?
Other articles by Dr. Cecchi: Diagnosing psychosis with word analysis.
More stories

A new supercomputing-powered weather model may ready us for Exascale

In the U.S. alone, extreme weather caused some 297 deaths and $53.5 billion in economic damage in 2016. Globally, natural disasters caused $175 billion in damage. It’s essential for governments, business and people to receive advance warning of wild weather in order to minimize its impact, yet today the information we get is limited. Current […]

Continue reading

DREAM Challenge results: Can machine learning help improve accuracy in breast cancer screening?

        Breast Cancer is the most common cancer in women. It is estimated that one out of eight women will be diagnosed with breast cancer in their lifetime. The good news is that 99 percent of women whose breast cancer was detected early (stage 1 or 0) survive beyond five years after […]

Continue reading

Computational Neuroscience

New Issue of the IBM Journal of Research and Development   Understanding the brain’s dynamics is of central importance to neuroscience. Our ability to observe, model, and infer from neuroscientific data the principles and mechanisms of brain dynamics determines our ability to understand the brain’s unusual cognitive and behavioral capabilities. Our guest editors, James Kozloski, […]

Continue reading