Chronic pain affects nearly one in five adults worldwide and remains one of the leading causes of disability. Unlike acute pain triggered by injury, chronic pain often arises spontaneously—without an obvious external cause—and fluctuates across minutes, hours, and days. Yet clinicians still rely largely on self-reported pain ratings, as there is currently no objective biomarker comparable to blood pressure or body temperature.

Now, a research team led by Associate Director WOO Choong-Wan at the Center for Neuroscience Imaging Research (CNIR) within the Institute for Basic Science (IBS), in collaboration with Professor CHO Sungkun’s team at Chungnam National University, has demonstrated that personalized brain-imaging models can decode fluctuations in spontaneous pain intensity in individuals with chronic pain.

To address the challenge of capturing internally generated pain, the researchers conducted an intensive longitudinal study in patients with fibromyalgia, a chronic pain disorder characterized by widespread spontaneous pain. Over more than half a year, participants underwent repeated functional magnetic resonance imaging (fMRI) sessions while continuously reporting their ongoing pain levels. fMRI measures changes in blood oxygenation across the whole brain, allowing researchers to examine patterns of communication between brain regions.

Using machine learning techniques applied to these densely sampled datasets, the team developed person-specific brain decoding models capable of predicting each participant’s moment-by-moment pain intensity. The models successfully tracked fluctuations in spontaneous pain across multiple timescales—from minute-level changes within a scan to differences across sessions and days.

Importantly, prediction performance improved substantially as more training data were included. Conventional data quantities, typically used in longitudinal neuroimaging studies, were insufficient to achieve reliable predictions. These findings highlight the importance of extensive within-person sampling when developing individualized brain-based biomarkers.

A key discovery of the study is that the neural patterns underlying pain differed markedly between individuals. The “pain connectome” identified in one participant did not generalize to another, and cross-testing between participants failed to produce meaningful predictions. This individual specificity suggests that chronic pain is represented through highly personalized brain network configurations, underscoring the limitations of one-size-fits-all biomarkers.

Unlike earlier personalized decoding studies that relied on recordings from a limited number of brain regions, the present study leveraged whole-brain functional connectivity, capturing distributed network interactions known to be involved in pain processing. The results demonstrate that spontaneous pain can be tracked using non-invasive neuroimaging, providing proof of principle for precision neuroimaging approaches in chronic pain research.

“The fact that pain cannot be seen adds to the suffering of patients with chronic pain,” said Dr. WOO Choong-Wan, Associate Director of IBS CNIR and senior author of the study. “Our findings show that precision neuroimaging may help evaluate invisible pain more objectively at the individual level.”

First author LEE Jae-Joong added, “Each participant exhibited a unique brain connectivity pattern associated with pain. Understanding these personalized neural signatures may eventually help guide precision approaches to pain assessment and treatment.”

Although the study involved a small number of participants and is not yet ready for clinical application, it establishes a methodological framework for developing patient-specific brain biomarkers. Future research involving larger and more diverse cohorts will be necessary to determine whether subtypes of chronic pain share common neural features or require entirely individualized solutions.

Figure 1. Pain intensity predicted by fMRI-based chronic pain marker
The horizontal axis represents the actual pain intensity reported by each participant, and the vertical axis represents the pain intensity predicted by the brain imaging marker, demonstrating the correspondence between actual and predicted pain. Different colors indicate different temporal units of prediction.

Figure 2. Regional feature importance of fMRI-based chronic pain marker
The color of each region represents the importance value, calculated as the decrease in prediction accuracy when that region is excluded from the marker, with the top five regions with the highest importance values additionally marked. The regions with high importance values differ across participants.