Yiyu Wang

All Publications

• Pharmacoepidemiologic Characterization of Cannabis Use and Symptomatology in Rheumatology using Natural Language Processing of Electronic Health Record Clinic Notes. The journal of pain Falasinnu, T., Le, N., Wang, Y., Alagappan, A., Walker, A., Park, T., Leung, J., Chaichian, Y., Weisman, M., Kenney, M., Irani, A., Bozkurt, S. 2025: 105633

  Abstract

  Up to 70% of patients with autoimmune rheumatic diseases (ARDs), including rheumatoid arthritis, psoriatic arthritis, and systemic lupus erythematosus, report moderate to severe pain despite controlled inflammation, driving interest in self-management including use of cannabis. We applied natural language processing (NLP) to 2.6 million electronic health record notes from 5,051 adults with ARDs seen at a tertiary health center. NLP classified cannabis documentation as current, past, or none and identified reasons (pain, sleep, anxiety, nausea, appetite). Classifiers achieved an F1 score of 0.85 for current versus past use and 0.83 for reasons, indicating a high level of accuracy. From 2004 to 2024, notes documenting current use rose from 0.1% to 1.1% (a 900% increase. Overall, 1,237 patients (24.5%) had ≥ 1 note of current use; prevalence was higher among Hispanic/Latino (30.1%) and Black (36.2%) patients than White (26.5%). Pain was the leading motive (37.9%), especially among Black (54.5%) and Hispanic/Latino (43.2%) patients, and women more often cited use for pain (39.4% vs. 33.0%) and sleep (16.4% vs. 11.6%) than men. Cannabis users had higher comorbidity indices, more emergency visits (2.1 vs. 1.3 per patient-year) and hospitalizations (1.4 vs. 0.9), and more opioid prescriptions (65% vs. 32.7%). These findings suggest rising cannabis use for ARD pain management and significant sociodemographic disparities, underscoring the need for prospective studies to assess outcomes and inform guidelines. PERSPECTIVE: This study demonstrates the feasibility of using natural language processing to extract real-world evidence on cannabis use in autoimmune rheumatic diseases. Findings reveal increasing documentation and sociodemographic disparities, underscoring the need for standardized recording and prospective studies to evaluate safety, effectiveness, and equitable access to cannabis-based symptom management.

  View details for DOI 10.1016/j.jpain.2025.105633

  View details for PubMedID 41354127

• Differential encoding of noxious heat and self-reported pain along corticospinal networks: a simultaneous spinal cord-brain fMRI study. bioRxiv : the preprint server for biology Pfyffer, D., Wang, Y., Kaptan, M., Fundaun, J., Dildine, T. C., Oliva, V., Indriolo, T., Skare, S., Sprenger, T., Lee, P. K., Truong, M., Weber, K. A., Glover, G. H., Law, C. S., Mackey, S. 2025

  Abstract

  Chronic pain poses a substantial public health burden. Elucidating how the healthy central nervous system (CNS) differentially encodes objective stimulus intensity and subjective experiences of pain perception may offer key insights into the central mechanisms contributing to chronic pain. Functional MRI (fMRI) combined with controlled noxious stimulation provides a powerful means to explore neural representations of nociception and pain perception. Here, we applied noxious heat at three intensities (46 °C, 47 °C, 48 °C, 8 trials each randomized) to the right forearm of 28 healthy women during simultaneous spinal cord-brain fMRI to investigate how distributed corticospinal activity and connectivity encode stimulus intensity and subjective pain. Activity increased with stimulus temperature across regions involved in pain processing-including somatosensory, motor, prefrontal, insular, and subcortical areas-as well as in the ipsilateral dorsal and ventral spinal cord. Spinal-brain functional connectivity was observed between the right dorsal horn and pain-related brain regions such as primary and secondary somatosensory cortex, insula, anterior cingulate cortex, thalamus, and periaqueductal gray, and was positively associated with individual pain ratings. Using representational similarity analysis (RSA), we found that multivoxel activation patterns in the brain and spinal cord, as well as corticospinal connectivity patterns, reliably tracked stimulus temperature, while only subsets of cortical regions (e.g., insula, sensorimotor, and frontal cortices) encoded subjective pain. Notably, spinal cord representations were primarily organized by stimulus temperature rather than perceived pain intensity. These findings demonstrate that simultaneous spinal cord-brain fMRI combined with multivariate modeling can identify sensory and perceptual components of nociceptive processing across the neuroaxis. Such approaches advance mechanistic understanding of pain and may inform the development of CNS-based biomarkers for chronic pain assessment and intervention.

  View details for DOI 10.1101/2025.10.24.684476

  View details for PubMedID 41280069

  View details for PubMedCentralID PMC12633297

• Big team science reveals promises and limitations of machine learning efforts to model physiological markers of affective experience ROYAL SOCIETY OPEN SCIENCE Coles, N. A., Perz, B., Behnke, M., Eichstaedt, J. C., Kim, S., Vu, T. N., Raman, C., Tejada, J., Huynh, V., Zhang, G., Cui, T., Podder, S., Chavda, R., Pandey, S., Upadhyay, A., Padilla-Buritica, J. I., Barrera Causil, C. J., Ji, L., Dollack, F., Kiyokawa, K., Liu, H., Perusquia-Hernandez, M., Uchiyama, H., Wei, X., Cao, H., Yang, Z., Iancarelli, A., Mcveigh, K., Wang, Y., Berwian, I. M., Chiu, J. C., Mirea, D., Nook, E. C., Vartiainen, H. I., Whiting, C., Cho, Y., Chow, S., Fisher, Z. F., Li, Y., Xiong, X., Shen, Y., Tagliazucchi, E., Bugnon, L. A., Ospina, R., Bruno, N. M., D'Amelio, T. A., Zamberlan, F., Mercado Diaz, L. R., Pinzon-Arenas, J. O., Posada-Quintero, H. F., Bilalpur, M., Hinduja, S., Marmolejo-Ramos, F., Canavan, S., Jivnani, L., Saganowski, S. 2025; 12 (6): 241778

  Abstract

  Researchers are increasingly using machine learning to study physiological markers of emotion. We evaluated the promises and limitations of this approach via a big team science competition. Twelve teams competed to predict self-reported affective experiences using a multi-modal set of peripheral nervous system measures. Models were trained and tested in multiple ways: with data divided by participants, targeted emotion, inductions, and time. In 100% of tests, teams outperformed baseline models that made random predictions. In 46% of tests, teams also outperformed baseline models that relied on the simple average of ratings from training datasets. More notably, results uncovered a methodological challenge: multiplicative constraints on generalizability. Inferences about the accuracy and theoretical implications of machine learning efforts depended not only on their architecture, but also how they were trained, tested, and evaluated. For example, some teams performed better when tested on observations from the same (vs. different) subjects seen during training. Such results could be interpreted as evidence against claims of universality. However, such conclusions would be premature because other teams exhibited the opposite pattern. Taken together, results illustrate how big team science can be leveraged to understand the promises and limitations of machine learning methods in affective science and beyond.

  View details for DOI 10.1098/rsos.241778

  View details for Web of Science ID 001514008500004

  View details for PubMedID 40568544

  View details for PubMedCentralID PMC12187420

• Neural Predictors of Fear Depend on the Situation. The Journal of neuroscience : the official journal of the Society for Neuroscience Wang, Y., Kragel, P. A., Satpute, A. B. 2024; 44 (46)

  Abstract

  The extent to which neural representations of fear experience depend on or generalize across the situational context has remained unclear. We systematically manipulated variation within and across three distinct fear-evocative situations including fear of heights, spiders, and social threats. Participants (n = 21; 10 females and 11 males) viewed ∼20 s clips depicting spiders, heights, or social encounters and rated fear after each video. Searchlight multivoxel pattern analysis was used to identify whether and which brain regions carry information that predicts fear experience and the degree to which the fear-predictive neural codes in these areas depend on or generalize across the situations. The overwhelming majority of brain regions carrying information about fear did so in a situation-dependent manner. These findings suggest that local neural representations of fear experience are unlikely to involve a singular pattern but rather a collection of multiple heterogeneous brain states.

  View details for DOI 10.1523/JNEUROSCI.0142-23.2024

  View details for PubMedID 39375037

  View details for PubMedCentralID PMC11561869

• A Computational Neural Model for Mapping Degenerate Neural Architectures. Neuroinformatics Khan, Z., Wang, Y., Sennesh, E., Dy, J., Ostadabbas, S., van de Meent, J. W., Hutchinson, J. B., Satpute, A. B. 2022; 20 (4): 965-979

  Abstract

  Degeneracy in biological systems refers to a many-to-one mapping between physical structures and their functional (including psychological) outcomes. Despite the ubiquity of the phenomenon, traditional analytical tools for modeling degeneracy in neuroscience are extremely limited. In this study, we generated synthetic datasets to describe three situations of degeneracy in fMRI data to demonstrate the limitations of the current univariate approach. We describe a novel computational approach for the analysis referred to as neural topographic factor analysis (NTFA). NTFA is designed to capture variations in neural activity across task conditions and participants. The advantage of this discovery-oriented approach is to reveal whether and how experimental trials and participants cluster into task conditions and participant groups. We applied NTFA on simulated data, revealing the appropriate degeneracy assumption in all three situations and demonstrating NTFA's utility in uncovering degeneracy. Lastly, we discussed the importance of testing degeneracy in fMRI data and the implications of applying NTFA to do so.

  View details for DOI 10.1007/s12021-022-09580-9

  View details for PubMedID 35349109

  View details for PubMedCentralID PMC9588472

• Neural Topographic Factor Analysis for fMRI Data Sennesh, E., Khan, Z., Wang, Y., Dy, J., Satpute, A., Hutchinson, J., van de Meent, J. edited by Larochelle, H., Ranzato, M., Hadsell, R., Balcan, M. F., Lin, H. NEURAL INFORMATION PROCESSING SYSTEMS (NIPS). 2020

  View details for Web of Science ID 001207696403013

• Neurocomputational mechanisms underlying motivated seeing. Nature human behaviour Leong, Y. C., Hughes, B. L., Wang, Y., Zaki, J. 2019

  Abstract

  People tend to believe that their perceptions are veridical representations of the world, but also commonly report perceiving what they want to see or hear. It remains unclear whether this reflects an actual change in what people perceive or merely a bias in their responding. Here we manipulated the percept that participants wanted to see as they performed a visual categorization task. Even though the reward-maximizing strategy was to perform the task accurately, the manipulation biased participants' perceptual judgements. Motivation increased neural activity selective for the motivationally relevant category, indicating a bias in participants' neural representation of the presented image. Using a drift diffusion model, we decomposed motivated seeing into response and perceptual components. Response bias was associated with anticipatory activity in the nucleus accumbens, whereas perceptual bias tracked category-selective neural activity. Our results provide a computational description of how the drive for reward leads to inaccurate representations of the world.

  View details for DOI 10.1038/s41562-019-0637-z

  View details for PubMedID 31263289