🎓 Academic
Xi Zhang
pic

Name: Xi Zhang

Affiliation: Liangzhu Laboratory

Business Email: oca-john@hotmail.com

Despite the widespread adoption of deep learning facilitated by general models, I remain firmly convinced that highly specialized models are irreplaceable in the realm of expertise. Drawing inspiration from biological knowledge and employing computational techniques, I strive for an effective integration of these disciplines in my research endeavors. Currently, my research focus resides in feature representation and feature fusion of time-series data, with the ultimate objective of developing robust analysis and prediction models for such signals. I am currently engaged in optical brain-computer interface research, primarily focusing on optical-based brain signal reading, decoding, optical signal writing, and encoding.

📰 Papers

  1. Inter-patient electrocardiogram quality assessment using cascaded domain adaptation.  (2026)
    Biomedical Signal Processing and Control, 113, 109033.  >>>  IF=4.9
    Li, H., Han, J., Zhang, X., Liu, Y., Xing, K., & Yang, H.*
  2. EMCNN: Fine-Grained Emotion Recognition based on PPG using Multi-scale Convolutional Neural Network.  (2025)
    Biomedical Signal Processing and Control, 105, 107594.  >>>  IF=4.9
    Han, J., Li, H., Zhang, X., Zhang, Y., & Yang, H.*
  3. A high altitude respiration and SpO2 dataset for assessing the human response to hypoxia.  (2024)
    Scientific Data, 11(1), 248.  >>>  IF=6.9
    Zhang, X., Zhang, Y.*, Si, Y., Gao, N., Zhang, H., & Yang, H.*
  4. Universal strategy for rapid design and analysis of gas detection peptide chips with positional preference.  (2024)
    Sensing and Bio-Sensing Research, 46, 100697.  >>>  IF=4.9
    Zhang, H., Zhang, X., Si, Y., Li, H., Han, J., Yang, C., & Yang, H.*
  5. A finer-grained high altitude EEG dataset for hypoxia levels assessment.  (2024)
    Scientific Data, 11(1), 1352.  >>>  IF=6.9
    Si, Y., Zhang, Y.*, Zhang, X.*, Liu, S., Zhang, H., & Yang, H.*
  6. Clinical knowledge-based ECG abnormalities detection using dual-view CNN-Transformer and external attention mechanism.  (2024)
    Computers in Biology and Medicine, 178, 108751.  >>>  IF=6.3
    Li, H., Han, J., Zhang, H., Zhang, X., Si, Y., Zhang, Y., Liu, Y., & Yang, H.*
  7. The spatiotemporal heterogeneity of the biophysical microenvironment during hematopoietic stem cell development: from embryo to adult.  (2023)
    Stem Cell Research & Therapy, 14(1), 251.  >>>  IF=7.3
    Shi, G., Zhang, P., Zhang, X., Li, J., Zheng, X., Yan, J., ... & Yang, H.*
  8. miRNA‐mediated macrophage behaviors responding to matrix stiffness and ox‐LDL.  (2020)
    Journal of Cellular Physiology, 235(9), 6139-6153.  >>>  IF=4.0
    Li, J., Wang, S., Li, Y., Zhang, N., Gribskov, M., Zhang, X., ... & Yang, H.*

📝 Review Experience

  1. Biomedical Signal Processing and Control
    Date: 2025  |  Role: Reviewer  |  Task ID: 0f1b8845-c81c-48b4-b414-54961ea759f7

📓 Research Plans

  1. Hypoxia State Prediction Model Based on Mirror Feature Curve (MFC) and Attractor Algorithm
    Developing a novel predictive framework that leverages mirror feature curves extracted from physiological signals combined with attractor-based dynamical systems analysis to forecast hypoxic episodes in high-altitude environments.
  2. Hypoxia State Prediction Based on MFC and Manifold Convolution Algorithm
    Integrating mirror feature curve representations with manifold-aware convolutional neural networks to capture the intrinsic geometric structure of physiological data for improved hypoxia classification and prediction.
  3. Hypoxia Event Modeling and Prediction Based on Graph Neural Networks
    Constructing graph-based representations of multi-channel physiological signals to model inter-channel dependencies and temporal dynamics for comprehensive hypoxia event detection and forecasting.
  4. All-Optical Closed-Loop Brain-Computer Interface Based on TCN and Attractor Algorithm
    Designing a fully optical brain-computer interface system utilizing temporal convolutional networks and attractor dynamics for real-time neural signal decoding and closed-loop neuromodulation.
  5. Dimensional Game Guide: Low-Dimensional Dynamics Modeling for Brain-Computer Interfaces
    Exploring low-dimensional manifold representations of neural population activity to develop interpretable and efficient decoding algorithms for next-generation brain-computer interfaces.
  6. Physics-Informed Neural Decoding: Synthetic-to-Real Domain Transfer
    Leveraging physics-informed neural networks to bridge the gap between synthetic training data and real neural recordings, enabling robust cross-domain generalization in neural decoding tasks.

🤝 Potential Collaboration Directions

  1. Optogenetic CAR-T Therapy (Opto-Immunotherapy)
    The major limitations of conventional CAR-T therapy are cytokine release syndrome (CRS) and off-target toxicity. By implementing NIR light-controlled CAR expression or IL-15/IL-12 release modulation, immediate light-off intervention becomes possible upon early CRS indicators (e.g., fever onset), providing a reversible safety mechanism with life-saving potential.
  2. Optogenetic Epigenetic Regulation (Opto-Epigenetics)
    Unlike direct DNA editing, epigenetic modulation offers reversible control—gene expression recovers upon light cessation. This enables researchers to fine-tune oncogene expression levels like adjusting a volume dial, observe tumor drug resistance responses, and identify dynamic equilibrium points. This approach aligns well with real-time deep learning training paradigms.