Rexis
  • Rexis: The L1 for DeSci
  • Key Challenges in BioDeSci
    • Overview
  • Data Level: Secure Biomedical Sharing
  • Model Level: Privacy-Preserving Fine-Tuning
  • Evaluation Level: Reproducible Computation
  • The Rexis Solution: Layer for DeSci
    • Overview
    • Data Level: Decentralized Biomedical Data Market
    • Model Level: Privacy-Preserving Training & Inference via Equivariant Encryption
    • Evaluation Level: Secure and Verifiable Biomedical Computation
  • Example BioDeSci Data and Models
    • Overview
  • Tabular Data
  • Biomedical Signals
  • Biological Sequences
  • Medical Imaging
  • Volumetric Medical Imaging
  • Spatial Omics Data
  • Tokenomics
    • $REX Overview
  • Links
    • rexis.io
  • Term of Use
  • Privacy Policy
  • Community
Powered by GitBook
On this page

Biomedical Signals

PreviousTabular DataNextBiological Sequences

Last updated 1 month ago

Biomedical Signals

Biomedical signals represent continuous measurements of physiological activity over time, offering dynamic insights into bodily functions. These signals are typically captured using specialized sensors and recording devices.

Electrocardiograms (ECG)

Electrocardiograms (ECGs) are a fundamental tool in cardiology. They record the electrical activity of the heart over time using electrodes placed on the skin. The resulting 1D signal reveals patterns related to:

  • Heart rhythm and rate

  • Conduction pathways

  • Abnormalities like arrhythmias, ischemia, and hypertrophy

Data Support

ECG data is commonly stored in formats like:

  • EDF (European Data Format)

  • MIT-BIH format

  • Simple text (e.g., CSV)

Our platform offers a secure repository and analysis platform for ECG data. Using Equivariant Encryption (EE), researchers and clinicians can store, share, and analyze ECGs securely—enabling collaborative large-scale ECG modeling while preserving patient privacy.

Model Support

The temporal structure of ECG data makes it ideal for transformer-based models. These models can:

  • Classify arrhythmias

  • Predict heart failure

  • Enable early diagnosis and continuous monitoring

Rexis supports secure evaluation and fine-tuning of ECG models using EE.

Photoplethysmograms (PPG)

Photoplethysmography (PPG) is a non-invasive optical technique used to detect changes in blood volume. PPG sensors shine light into the skin and measure the reflected or transmitted signal, capturing the pulse waveform.

PPG is commonly used in:

  • Smartwatches

  • Fitness trackers

  • Wearable health monitors

Data Support

PPG signals are typically stored in:

  • EDF

  • Vendor-specific proprietary formats

Our platform provides a secure and scalable infrastructure for analyzing large-scale PPG datasets collected from wearables. EE ensures privacy by enabling encrypted storage and secure distributed processing.

Model Support

Foundation models are increasingly applied to PPG signals for multi-task learning. They support:

  • Age/gender estimation

  • Vital sign prediction

  • Stress detection

Rexis enables encrypted evaluation and training for PPG-based health models.

Electroencephalograms (EEG)

Electroencephalograms (EEGs) record the brain’s electrical activity using electrodes on the scalp. EEG signals are key in both clinical diagnostics and neuroscience research.

Used in diagnosing:

  • Epilepsy

  • Sleep disorders

  • Encephalopathies

EEG is also used to analyze:

  • Cognitive states

  • Attention and workload

  • Brain-computer interfaces (BCIs)

Data Support

EEG data is often stored in:

  • EDF

  • Manufacturer-specific binary formats

Our platform offers a secure, collaborative environment for EEG data sharing and analysis. With EE, sensitive neurological data is protected during model training and evaluation—supporting reproducibility and privacy.

Model Support

Transformer-based models have shown strong performance in EEG-based tasks, such as:

  • Seizure detection

  • Sleep staging

  • Cognitive state decoding

Rexis supports secure fine-tuning and inference on encrypted EEG data.

Wearable Sensor Data

Wearable devices now monitor a variety of physiological and behavioral metrics in real-time. Common sensor types include:

  • Accelerometers (motion)

  • Gyroscopes (rotation)

  • Magnetometers (magnetic field)

  • Thermometers

  • Electrodermal Activity (EDA) sensors

These generate continuous 1D time series data used for:

  • Activity recognition

  • Sleep monitoring

  • Stress and HRV analysis

  • Real-world health tracking

Data Support

Sensor data is often stored in:

  • CSV

  • JSON

  • Proprietary wearable SDK formats

EE ensures the privacy and secure processing of this high-frequency data, enabling encrypted model training and inference without compromising user privacy.

Model Support

Foundation models trained on wearable data enable:

  • Personalized health modeling

  • Remote patient monitoring

  • Generalizable embeddings for human activity

Our platform enables encrypted evaluation and fine-tuning of wearable models.

: A transformer-based foundation model trained on large ECG datasets. It supports tasks like arrhythmia classification and heart failure prediction.

: A multi-task model that processes PPG signals to estimate physiological attributes such as age and gender.

: A transformer-based EEG foundation model designed for seizure detection, sleep stage classification, and other downstream neuroscience tasks.

: A self-supervised model for human motion analysis using wearable sensor data. Supports applications like activity recognition and behavior modeling.

ECG-FM
PaPaGei
LaBraM
Wearable-SSL