FACE project: AI, cloud and edge for more precise ECG analyses
We spoke to Maximilian Zillekens about the innovative FACE project, which aims to revolutionize long-term ECG diagnostics with the help of AI as well as cloud and edge computing. Zillekens, software engineer at GETEMED Medizin- und Informationstechnik AG, explains how these technologies can improve the efficiency and accuracy of ECG analysis while overcoming challenges such as data protection and interoperability. Read more in the following interview
About
"My name is Maximilian Zillekens, I am 30 years old and have been working as a software engineer at GETEMED Medizin- und Informationstechnik AG for five years. I mainly work in the development of AI algorithms for ECG analysis.
What excites me about my work is transferring modern technologies into applications that can then be used in everyday medical practice. I have been fascinated by the combination of software and medicine since my studies.
In software development itself, I like the structured way of solving complex tasks by breaking down large projects into smaller sub-problems."
Picture (c) Maximilian Zillekens, LinkedIn
SEMDATEX: Mr. Zillekens, what is the FACE project and what are its main objectives? Who are the main beneficiaries of this project and what innovative approaches make it particularly valuable for the healthcare sector?
Maximilian Zillekens: The FACE project aims to develop a modern, AI-supported analysis platform for evaluating long-term ECGs. We are relying on a combination of cloud and edge computing for this. This means that an initial analysis is carried out in the doctor's practice on the practice computer (edge). In addition to the medical parameters, the AI model also calculates the uncertainty of these results so that only complex cases are uploaded to the cloud for more in-depth analysis.
What are the specific advantages of the FACE infrastructure for analyzing ECG data compared to previous approaches? Why is the combination of edge and cloud computing so important in this context?
This combination enables resource-saving and efficient analysis. The selective transfer of long-term ECG recordings minimizes transport and calculation costs on cloud systems. At the same time, the quality of the analysis is ensured, as difficult ECG recordings are analyzed in detail in the cloud.
What challenges are there in the technical implementation of the FACE project, particularly in terms of interoperability, data protection and security? How will the project be funded and supported?
To ensure interoperability, we are planning to integrate common interfaces to PVS and HIS systems. In terms of data protection, we are focusing on avoiding the transfer of personal data to the cloud and processing sensitive data locally. At the same time, we attach great importance to implementing established security standards and are planning penetration tests by an independent company. This project is funded by the German Federal Ministry for Economic Affairs and Climate on the basis of a decision by the German Bundestag and financed by the European Union.
About the project
FACE stands for "Research on the application of cloud and edge computing for efficient AI-supported ECG analysis" and is developing an infrastructure for analyzing ECG data and other health data. A system architecture for edge and cloud data processing is being created to read out ECGs, assign them to patients and analyze them in a 2-step process. The ECG interface (GETEMED) enables the data to be transferred to an analysis platform (SEMDATEX). This presents the results to doctors and is suitable for use on site and in the cloud. The ECG data is stored in the data warehouse and transferred to the cloud for further analysis if required.
Could you give us an overview of the various consortium partners in the FACE project? How do these partners contribute to the overall success of the project with their specific expertise?
The DHZC (German Heart Center of the Charité) and the Evangelisches Diakonissenhaus Berlin-Teltow-Lehnin are represented as medical partners in the project. These partners contribute medical expertise in order to develop the platform in a user-centered manner, analyze ECGs and test the system in everyday clinical practice. In addition to GETEMED, Charité's BIH and the University of Siegen are also involved in the development of the AI algorithms. SEMDATEX and BIOTRONIK are developing the analysis platform and communicating extensively with future users in order to be able to develop iteratively and agilely close to the user. This interdisciplinary consortium brings together a wide range of expertise, which makes the exchange and cooperation with all partners particularly valuable for us.
What tasks does GETEMED take on as a consortium partner in the FACE project? To what extent does GETEMED contribute its expertise in medical technology, particularly in the integration of ECG recordings?
GETEMED is the consortium leader and is developing AI models that are used as cloud-based solutions as part of the FACE project. We are also working on web-based modules to enable modern ECG analysis. With over 35 years of experience in ECG analysis, we bring in-depth domain knowledge, which we combine with state-of-the-art AI-based algorithms. As consortium leader, we also coordinate the overall project and ensure close cooperation between all partners.
What exactly are your tasks within the FACE project? What technical aspects or challenges in the project are part of your area of responsibility?
As part of the FACE project, I am responsible for the development of AI models for ECG analysis. My focus is on combining domain knowledge from ECG analysis with state-of-the-art AI methods. A major technical challenge here is the long iteration time associated with training large AI models until new results can be evaluated. When developing an app, you can quickly check whether the new button works or the color is displayed correctly, for example. With AI algorithms, it takes longer to check whether a small model adjustment actually leads to better results. My team is also responsible for the development and integration of the ECG analysis modules. As part of this, I am actively involved in architecture planning, code reviews and discussions and decision-making on key design issues.
Can you go into more detail about the technological implementation? Which interfaces, platforms or AI models are used to enable effective ECG data analysis?
We rely on state-of-the-art architectures for our AI models, which are based on transformer layers, for example. At the same time, we are working intensively on automating large parts of our development cycle. Our vision is to automatically retrain, evaluate, document and deploy models as soon as new, annotated data is available. Although this is certainly still a long way off, the gradual automation of each individual process is helping us to make steady progress.
What do you hope for the future of the FACE project? Are there any plans to turn the technologies developed in the project into concrete products and what opportunities do you see for their market potential?
Once the project is completed, we aim to transform the technologies developed into market-ready medical products. We see great potential for these solutions in the healthcare sector, as they can make ECG analysis safe, efficient and significantly faster. We anticipate strong market acceptance, particularly in view of the increasing demand for time-saving solutions for long-term ECG analysis. In the long term, the technologies could also be transferred to other areas of medical diagnostics, opening up further exciting opportunities.