AIQNET – Medical Data Ecosystem
Driving medical progress with AI
Although medical data is urgently needed by medical device manufacturers and clinics for scientific studies and to meet regulatory requirements, there are many hurdles to overcome in collecting and using it. In most cases, the data is distributed across non-interoperable systems or is available in different formats. The AIQNET project is developing a digital ecosystem that enables the use of medical data across sectors and in compliance with data protection regulations. The acquisition and analysis of the data will be largely automated with AI.
In order to be able to market the project internationally and to emphasise the holistic approach, it was renamed from KIKS ("Artificial Intelligence for Clinical Trials") to AIQNET.
Market perspective and product promise
The AIQNET digital ecosystem aims to structure and make available medical data through software applications - modelled on the app stores on mobile devices. Various parties benefit from this: Automated data collection with AI applications can eliminate time-consuming tasks such as transferring paper-based information into IT systems, leaving more time for treatment. For medical technology companies, access to product-related data makes it much easier to fulfil their legal obligation for ongoing product monitoring and to conduct clinical trials. Software providers can use the provided infrastructure and access to medical data to develop data-driven applications in a short time. For the operation of the ecosystem, a revenue share is levied on the providers of the applications.
Consortium
Raylytic GmbH, Berlin Cert - Prüf- und Zertifizierstelle für Medizinprodukte GmbH, BioLago e.V., BioRegio STERN Management GmbH, BIOTRONIK SE & Co. KG, Charité - Universitätsmedizin Berlin, Eberhard Karls Universität Tübingen, ExB Research & Development GmbH, HWI pharma services GmbH, inomed Medizintechnik GmbH, Aesculap AG, MedicalMountains GmbH, Universitätsklinikum Magdeburg A.ö.R., TZM GmbH, Universitätsklinikum Jena, Universität Leipzig
Challenge and innovation
The costs of clinical trials are very high. However, manufacturers of medical devices are permanently dependent on up-to-date medical data for the development and control of their products. Doctors also need them, e.g. for diagnosis or for their treatment decisions, as do healthcare providers and hospital operators. However, the lack of specialists in industry and hospitals, legal uncertainties and isolated IT systems with low interoperability have so far hindered a comprehensive collection of medical data. In the AIQNET project, industry partners are working together with hospitals for the first time to process medical data in such a way that it can be used for different purposes in compliance with data protection laws.
Solution
Patient data is predominantly available as continuous text (e.g. laboratory reports) or as image information (e.g. X-ray images). This unstructured data must first be converted into a structured form for data processing. The project agrees on a common standard for data description and at the same time creates the possibility to integrate other standards so that the data can be exchanged. In addition, interfaces to the various IT systems of hospitals and medical device manufacturers are being developed.
The project uses two AI technologies to automate the analysis: Convolutional Neural Networks (CNN) are used to analyse the image data, which are trained through "supervised learning" to reliably interpret image data such as X-ray images. In so-called Natural Language Processing (NLP), methods and techniques from linguistics, computer science and artificial intelligence are combined so that continuous texts can be processed by machines. Finally, the data is translated into a format that enables the exchange of data across sector boundaries.
For both texts and images, features that allow conclusions to be drawn about individuals will be removed from the data sets. The project is also developing a procedure that can be used to determine the probability of identifying people by combining individual data. Data can thus be reduced to its actual information content and used for tasks in research, industry, diagnosis and treatment. For the legally sound and transparent use of data, patients are empowered to determine and control the eventual use of their data. The ecosystem thus ensures that providers and users of the data always act within the framework of the applicable data protection guidelines.
Use Cases
At the launch of the marketplace, several relevant applications will be available on which the consortium partners are already working. The applications demonstrate how clinical data can help improve healthcare and patient care.
Automated wear analysis of hip implants
Until now, radiologists have had to use many X-ray images to check the condition of the implant. AIQNET's image recognition automates this routine task, leaving the doctor more time for the patient.
Monitoring the safety of medicines
Medical device manufacturers are obliged to set up a so-called "vigilance system" for the continuous monitoring of any side effects that may occur with their products. AIQNET's AI-based solutions largely automate this labour-intensive task while ensuring compliance with data protection.
Providing market monitoring data for cardiology products
Manufacturers of medical products such as heart valves are required by the European Medical Device Regulation (MDR) to compare the performance of their products with competing products. In AIQNET's marketplace, there will be applications that provide representative and traceable data on the performance of cardiology products.
