Bayesian inference with flexible electronics for biomedical applications

An EIC Pathfinder project.

Overview

BAYFLEX is developing organic environmentally sustainable electronics for probabilistic computing on flexible substrates. The project will validate this approach through the detection and classification of electrophysiological signals, introducing a new class of low-power, eco-friendly wearables that can sense data and make decisions.

Discover the project Sustainalibity Technological impact

Outreach video : comming soon.

Who are we ?

BAYFLEX is a European partnership between different 3 academic instituions (France, Spain, Germany), a research and technology organization (France) and a neuro-tech (Spain).

About us

Organisation

All

Stochastic computing

In BAYFLEX we introduce neuron based stochastic computing circuits. Time encoded streams are realized by spiking neuron circuits and used to encode probabilities. Classification is realized using Bayesian inference.

Device Modeling

Modeling devices is first done in TCAD to understand the most important physics. Compact models compatible with circuit simulations are then developed. Challenges encountered in BAYFLEX include modeling the non-linearities in sub-threshold regime and time-dependent characteristics .

Organic Electrochemical Transistors (OECTs)

OECT technology is developed for active sensor electrodes, to measure and amplify electro-physiological signals. In addition, the hysteresis in OECTs is used to implement a Muller c-element (MCE) and perform Bayesian Inference.

OTFTs and Integration

Low threshold OTFT circuits are developed for transforming features into probabilities. The three technologies (OECT active sensors, OTFT circuits, OECT classification circuits) are integrated into a single substrate.

Applications

BAYFLEX aims to detect sleep stages in electrophysiological data in order to enable auditory stimulation that can improve sleep. Initial tests on epilepsy EEG and ECG data demonstrate the viability of the concepts.

LCA & Sustainability

A process of record was first established for the BAYFLEX technologies. An initial life cycle (cradle to use) assessment shows that the largest green house gas emissions are due to photolithography processes.