Background & Aims
Pain is a complex and subjective experience and exhibits individual variability in its brain representation [1,2]. This poses a challenge in understanding the neurobiological mechanisms of pain at the group level. In efforts to develop clinically useful pain biomarkers, fMRI-based pain biomarkers have been developed using predictive modeling approaches [3, 4]. Despite demonstrating general neural representations of pain and predictive capacities across individuals, their clinical potential remains elusive. The precision neuroimaging approach involves dense sampling of single individuals and holds promise for understanding neurobiological mechanisms of pain and obtaining reliable within-individual measures [5, 6]. In this study, we developed a personalized pain predictive model using densely sampled data from a single participant. Using the personalized model, we aimed to understand the clinical utility of fMRI-based pain biomarkers and the differences from the population-level model.
Methods
For the individual dataset, one healthy, right-handed participant completed 33 sessions of the fMRI experiment. The population dataset comprised 124 participants undergoing a single session of the same experiment. During the fMRI scan, participants experienced a total of 96 thermal pain stimuli, ranging from 45 to 47.5? in 0.5? increments, and rated the magnitude of pain. To develop multivariate pattern-based predictive models, we randomly divided the single-trial fMRI datasets into training, validation, and test sets based on sessions. We trained each model to predict pain ratings using principal component regression (PCR) with leave-one-session-out cross-validation and tested the models on the independent test datasets. To compare the two models, we performed searchlight-based pattern similarity and virtual lesion analyses. Lastly, we combined the three models into an ensemble model and quantified the unique and shared variance explained by each pain predictive model.
Results
We tested the personalized pain predictive model on an independent within-individual dataset, and it showed a high predictive performance (r = 0.786, R-squared = 0.617). This model outperformed the population-level model (r = 0.459) in predicting pain within the individual. The weight patterns of the two models were highly similar in the insula and supplementary motor cortex, while the lateral thalamus and lateral prefrontal cortex showed low similarity. Finally, we employed the ensemble model framework with the experimental parameter model along with the two brain models. The experimental parameter model estimates the individual’s pain ratings based on the experimental information without neural information. The ensemble model showed that the individual brain model has a unique variance (18.5% [0.127/0.685]) that is not explained by the population brain or the experimental parameter models.
Conclusions
Our results showed that the personalized pain predictive model outperformed the population-level model in predicting pain within an individual. While both the personalized and population-level models identified pain-related brain regions as important, the personalized model placed greater importance on higher-order areas. The personalized brain model had a unique variance in pain experience over and above the population brain or the experimental parameter models. These findings highlight the importance of developing personalized neuroimaging-based pain biomarkers, supporting personalized pain assessment and treatment.
References
[1] Fillingim, R. B. (2017). Individual differences in pain: understanding the mosaic that makes pain personal. Pain, 158(Suppl 1), S11.
[2] Kohoutová, L., Atlas, L. Y., Büchel, C., Buhle, J. T., Geuter, S., Jepma, M., … & Woo, C. W. (2022). Individual variability in brain representations of pain. Nature neuroscience, 25(6), 749-759.
[3] Wager, T. D., Atlas, L. Y., Lindquist, M. A., Roy, M., Woo, C. W., & Kross, E. (2013). An fMRI-based neurologic signature of physical pain. New England Journal of Medicine, 368(15), 1388-1397.
[4] Woo, C. W., Schmidt, L., Krishnan, A., Jepma, M., Roy, M., Lindquist, M. A., … & Wager, T. D. (2017). Quantifying cerebral contributions to pain beyond nociception. Nature communications, 8(1), 14211.
[5] Gordon, E. M., Laumann, T. O., Gilmore, A. W., Newbold, D. J., Greene, D. J., Berg, J. J., … & Dosenbach, N. U. (2017). Precision functional mapping of individual human brains. Neuron, 95(4), 791-807.
[6] Kraus, B., Zinbarg, R., Braga, R. M., Nusslock, R., Mittal, V. A., & Gratton, C. (2023). Insights from Personalized Models of Brain and Behavior for Identifying Biomarkers in Psychiatry. Neuroscience & Biobehavioral Reviews, 105259.
Presenting Author
Youngeun Park
Poster Authors
Topics
- Pain Imaging