Skip to content
DraconiticResearch
Search Sources API
186,123 studies31 sourcesUpdated 46 min ago
Peer-reviewedCC BY-NCScience Advances

Biophysically constrained learning from healthy neural priors improves motor decoding in pathological cortical networks

  • Yihan Wu1,2
  • Tao Chang3
  • Siliang Chen3
  • Yuxin Quan3
  • Qiang Sun4
  • Marc M. Van Hulle4
  • Yanhui Liu3
  • Ning Jiang1,2
  • Dario Farina5
  • Yuan Yang3
  • Qing Mao3
  • Jiayuan He1,2
5 affiliations
  1. 1National Clinical Research Center for Geriatric, West China Hospital of Sichuan University, Chengdu 610017, Sichuan, China.
  2. 2Med-X Center for Manufacturing, Sichuan University, Chengdu 610017, Sichuan, China.
  3. 3Department of Neurosurgery, West China Hospital of Sichuan University, No. 37 Guo Xue Xiang, Chengdu 610041, China.
  4. 4Laboratory for Neuro- and Psychophysiology, Department of Neurosciences, KU Leuven, Leuven 3000, Belgium.
  5. 5Department of Bioengineering, Imperial College of Science, Technology and Medicine, London SW7 2AZ, UK.
Read full textPDFDOICite

Abstract

High-performance brain-computer interfaces (BCIs) rely on stable neural representations, yet brain functional impairments inherently disrupt canonical encoding, creating a fundamental decoding paradox. Conventional data-driven approaches struggle to bridge this gap because of pathological data scarcity and heterogeneity. Here, we propose to leverage healthy neural priors and introduce Pathology-Aware Consensus Training (PACT), a paradigm embedding biophysical constraints into healthy neural recordings to guide neural decoding. In a gesture classification task using high-density electrocorticography from patients with gliomas infiltrating the motor cortex, PACT improved the accuracy by more than 10%, effectively restoring the distorted semantic geometry of motor commands and realigning them with healthy canonical representations (correlation increase: r = 0.66 to r = 0.94). Mechanistically, PACT induced an emergent "inverse weighting" strategy: The model autonomously suppresses high-power pathological noise while reanchoring spatial attention to functional compensatory regions. This study provides a proof of concept for improving decoding in compromised networks by harnessing healthy neural dynamics, laying the theoretical foundation for future BCI deployments in clinical populations.

Related

  1. Deep-learning-based fMRI decoding of real-world size for hand-held objects
    Cui Lang, Kosuke Miyoshi, Masaki Takeda
    Scientific ReportsPeer-reviewedCC BY-NC-ND 4.0Full textNeuroscienceCrossref
  2. Decoding Clade-Specific bNAb-HIV-1 Env Interactions Using Machine Learning
    Y. Kumar, S.-T. Lin, N. M. Dixit +1
    bioRxivPreprintAll rights reservedFull textMicrobiologybioRxiv
  3. SpectroVQ: Noise-Aware Compression of Proteomics Data via Vector-Quantized Deep Learning improves MS/MS data storage and Peptide Identification
    H. Lam, J. H. W. Li, A. Hoque
    bioRxivPreprintAll rights reservedFull textBioinformatics & computational biologybioRxiv
  4. Clinical-Anchored Incremental Survival Learning: A Stability-Constrained Framework for Molecularly Enhanced Survival Prediction in Lung Adenocarcinoma
    P. Shringarpure, S. Divakar, S. Khobragade
    bioRxivPreprintAll rights reservedFull textBioinformatics & computational biologybioRxiv
  5. Decoding Extrahepatic Targeting of Lipid Nanoparticles with Interpretable Machine Learning
    Asal Mehradfar, Mohammad Shahab Sepehri, Owen Antholine +4
    arXivPreprintarXiv licenceFull textBioinformatics & computational biologyarXiv

Cite

BibTeXRIS
BibTeX
@article{wu2026biophysically,
  title = {{Biophysically constrained learning from healthy neural priors improves motor decoding in pathological cortical networks}},
  author = {Yihan Wu and Tao Chang and Siliang Chen and Yuxin Quan and Qiang Sun and Marc M. Van Hulle and Yanhui Liu and Ning Jiang and Dario Farina and Yuan Yang and Qing Mao and Jiayuan He},
  journal = {Science Advances},
  year = {2026},
  month = oct,
  doi = {10.1126/sciadv.adx5564},
  url = {https://www.science.org/doi/10.1126/sciadv.adx5564},
  note = {CC BY-NC},
}
RIS
TY  - JOUR
TI  - Biophysically constrained learning from healthy neural priors improves motor decoding in pathological cortical networks
AU  - Yihan Wu
AU  - Tao Chang
AU  - Siliang Chen
AU  - Yuxin Quan
AU  - Qiang Sun
AU  - Marc M. Van Hulle
AU  - Yanhui Liu
AU  - Ning Jiang
AU  - Dario Farina
AU  - Yuan Yang
AU  - Qing Mao
AU  - Jiayuan He
PY  - 2026
DA  - 2026/10/23
JO  - Science Advances
DO  - 10.1126/sciadv.adx5564
UR  - https://www.science.org/doi/10.1126/sciadv.adx5564
AB  - High-performance brain-computer interfaces (BCIs) rely on stable neural representations, yet brain functional impairments inherently disrupt canonical encoding, creating a fundamental decoding paradox. Conventional data-driven approaches struggle to bridge this gap because of pathological data scarcity and heterogeneity. Here, we propose to leverage healthy neural priors and introduce Pathology-Aware Consensus Training (PACT), a paradigm embedding biophysical constraints into healthy neural recordings to guide neural decoding. In a gesture classification task using high-density electrocorticography from patients with gliomas infiltrating the motor cortex, PACT improved the accuracy by more than 10%, effectively restoring the distorted semantic geometry of motor commands and realigning them with healthy canonical representations (correlation increase: r = 0.66 to r = 0.94). Mechanistically, PACT induced an emergent "inverse weighting" strategy: The model autonomously suppresses high-power pathological noise while reanchoring spatial attention to functional compensatory regions. This study provides a proof of concept for improving decoding in compromised networks by harnessing healthy neural dynamics, laying the theoretical foundation for future BCI deployments in clinical populations.
SN  - 2375-2548
ER  -