Published on 26 October 2023
The European BAYFLEX project aims to develop flexible, intelligent sensors capable of continuously collecting and processing biological signals before their on-chip classification. This ambitious project adds value by using "green" organic technologies, which require less energy to manufacture and use.
The application of digital technologies to healthcare has opened up a whole new world of screening, monitoring and treatment opportunities. However, such services are often expensive, or even inaccessible. The development of affordable technological solutions is therefore a key factor in the democratization of healthcare.
This issue is central to the BAYFLEX project as part of the Horizon Europe program’s EIC Pathfinder Open 2022 call for proposals. The project is seeking to develop a low-cost, flexible electronics device that uses artificial intelligence (AI) to track physiological signals. It could be used to monitor the brain’s electrophysiological activities for the detection, or even pre-detection, of abnormal activities. Target applications include the treatment of epilepsy and neurodegenerative diseases such as Alzheimer’s.
EIC Pathfinder is a highly competitive call for proposals that supports high-risk, high-potential proposals for the development of radically innovative technologies. BAYFLEX builds on the lessons learned from BAYOEN, an initial bilateral project involving the French National Research Agency (ANR) and the German Federal Ministry of Education and Research, which focused on the computation and classification module for this intelligent device. Launched on April 1st, 2023, BAYFLEX plans to cover the entire device production process within a three-year period: from capturing signals in the body to signal recovery, processing and transmission to the clinical world. Mines Saint-Étienne is participating in this ambitious program, which stands out for its resolutely “green” approach incorporating more energy-efficient technologies throughout the innovation chain.
The device envisaged by BAYFLEX consists of wearable sensors that use AI to detect anomalies. These sensors, made from flexible organic materials that interface with the human body, pick up an electrophysiological signal, such as an electrocardiogram (ECG). This signal is first processed and then classified using circuits. These circuits are described as “neuromorphic” because they are based on an AI method inspired by the functioning of biological neurons: Spiking Neural Networks (SNN).
Today, the classification of processed signals, which enables the detection of anomalies, is carried out by computer. That is why this type of device is still very energy-intensive. For BAYFLEX, however, classification will be carried out “on chip” using embedded organic transistors, which means that all the intelligence unfolds directly within the sensor.
The BAYFLEX device consists of sensors that process physiological signals from the human body, converting them into probabilized signals that can be used to detect anomalies.
The optimal assembly of all the elements on a single carrier is the subject of an entire section of the project. Based on the key parameters for each module, a team models and simulates the design that is identified as being the best and most energy-efficient. Identical production – using the same processes and the same organic materials – should ensure the compatibility of all components.
“The use of devices for on-chip classification has already been envisaged in ‘traditional’ silicon technologies. BAYFLEX offers the added value of a technology that will enable AI-based classification using environmentally-friendly, low-energy circuits,” maintains Esma Ismailova, a researcher at Mines Saint-Étienne specializing in wearable organic electronics for healthcare.
The energy bill for the silicon-based technologies currently in use is very high. Organic components are much less energy-intensive to produce, they require fewer chemical, toxic or corrosive pollutants, and in use, organic transistors require less energy to perform the same calculations. At the end of their life, these materials can also be recycled or reused, using less energy and limiting the generation of waste electrical and electronic equipment (WEEE, better known as e-waste). “These devices are made from carbon-based materials: there are no precious materials or costly, energy-intensive processes, so in the long term, they should be cheaper to produce than silicon-based devices,” adds the researcher.
BAYFLEX, like its forerunner BAYOEN, is the brainchild of Laurie Calvet, a CNRS researcher specializing in device physics. She met Esma Ismailova during a symposium organized by the Organic Electronics for the New Era (OERA) Research Group in Paris. “I was presenting my work on electrodes capable of recovering physiological signals and transforming them into electrical signals for health monitoring,” explains the Mines Saint-Étienne researcher. “On her first project, Laurie had already worked on the computational electronics part, but she lacked the interface link with the human body for this new project. That’s how she came to ask me to join the team.”
For Esma Ismailova, the challenge is to improve signal capture from patients equipped with a portable system. “Medical practices and healthcare institutions have robust technologies for collecting physiological signals, but outside this clinical setting, it gets more complex,” she notes. “The signal recorded by a portable sensor often comes out very noisy and very weak.” As a result, processing it requires more technology – a computer – and notably more energy to run the algorithms. To remedy this problem, the researcher is investigating combinations of passive and active electrodes that can capture and transmit the signal more robustly, thereby improving its quality.
For these electrodes, Esma Ismailova also uses organic materials based on conductive polymers. Their specificity is the ability to conduct both electrical and ionic signals, enabling the translation of a biological signal into an electrical one. Conductive polymers are also more compatible with the human body, as they are more flexible and mechanically adaptive, and therefore ergonomic.
On the one hand, the passive electrode is an electronic monitoring device: it records the physiological signal in the form of a potential variation, but does not transform it. Processing is always carried out by computer. On the other hand, the recorded signal is clean, with few artifacts and does not need to be excessively purified to recover critical information, which means simpler, less energy-intensive processing.
Hence the heightened interest in combining this electrode with an organic electrochemical transistor (OECT) to create an active electrode. Developed at the Bioelectronics Department at Mines Saint-Étienne, this innovative electrode is capable of recording potential, amplifying it and translating it into an electrical signal. OECTs are highly efficient at translating biological signals, and less sensitive to the effects of environmental noise (movement, electromagnetic fields, etc.), which once again means that a clean, high-quality, conditioned and compatible signal can be recovered, to facilitate on-chip computation.
The BAYFLEX project includes a dissemination component targeting a variety of audiences: scientists through publications in journals and conferences, manufacturers and end users through demonstrations, and the general public. For this last target, the consortium is intending to use the mediation expertise of La Rotonde, the Center for Scientific, Technical and Industrial Culture (CCSTI) at Mines Saint-Étienne.
By organizing a workshop aimed at non-scientific communities, the researchers involved in the project are hoping to effectively convey the many key concepts covered by this multidisciplinary project. “BAYFLEX is all about electrophysiological sensors, organic materials, neuromorphics, bioelectronics, physical modeling and computational processing… We feel that organizing a fun activity is the best way to communicate on this project, without generating misunderstandings or misinterpretations. The message we really want to convey is that today, we are capable of developing electronics that serve society, and that are greener, more energy-efficient and accessible to as many people as possible“, sums up Esma Ismailova.
The BAYFLEX project, standing for BAYesian Inference with FLEXible electronics for biomedical Applications, was launched on April 1st, 2023 for a three-year period. The six participants in the consortium – Bitbrain & Rovira i Virgil University (Spain), the CEA, CNRS & Mines Saint-Étienne (France), and the Technical University of Dresden (Germany) – bring together interdisciplinary expertise in emerging device modeling, biologically inspired circuit design, machine learning involving electrophysiological data, and OTFT and OECT fabrication. For more information, visit the project website: https://bayflex.cnrs.fr/
Swoop Energy gives a second life to electric vehicle batteries