Differential diagnosis

Practical example and current challenges

• Differential diagnosis is a crucial clinical process in which, right from the outset, decisions are made as to what steps should be taken in the diagnosis and treatment process. In most cases, medical staff has neither the time nor the resources to use the wealth of potentially useful information held in hospital systems, which could help them make a more targeted diagnosis.
• However, by using information from a patient’s medical records or physical examination right from the first anamnesis, as well as similar cohorts in record archives, critical diagnoses could be detected more quickly, and rare diseases were diagnosed in a more targeted manner. Besides, suggestions can be given for further medical measures to avoid unnecessary procedures.
• In order to improve the search and the selection of suitable cohorts, and to gain targeted knowledge/promote the exchange of knowledge on rare diseases, in particular, a comprehensive database and a common platform are needed.
• A major challenge in creating models to be used cross-organization is using the diverse stocks of data in the healthcare system to improve artificial intelligence, without harming the personal rights of patients.
• To date, there is little expertise in hospitals on neural medical history for latent neuronal patient representations. Such records can, in particular, integrate information from multimodal patient data based on text, lab results, medical technology, and images exceptionally well. Often, there is a lack of experienced team structures for industrial-scale production of medical data products analogous to those in place in comparable companies in the platform economy.

What added value does the "GAIA-X project" offer?

• According to a recent study, the market for clinical diagnostic support for Germany totals approximately €1.4 billion. As proposed in this use case, the market for electronic health records is worth even more than €6.4 billion. This use case is helping Germany secure its autonomy in this important AI technology.
• Creating a distributed data space for the exchange of data and algorithms between hospitals leads to better early detection when AI-based differential diagnosis is used, especially for rare diseases. Medical expertise from specialized centers in university hospitals could be made more accessible to rural regions, which could reduce the time patients have to suffer and wait until diagnosis, especially for rare medical problems.
• GAIA-X establishes both the necessary infrastructure as well as providing access to the latest AI analysis methods. Through the shared infrastructure, GAIA-X supports the re-use of uniform patient records in the hospital and other numerous applications, even outside of the individual organization.
• AI-based differential diagnosis via GAIA-X leads to a reduction in morbidity and mortality based on case-specific diagnostics, which in turn reduces costs.
• State-of-the-art security concepts are utilized that enable critical personal (health) data to be used in a secure, transparent, and useful manner.

Use Case Team

• Prof. Dr-Ing. habil. Alexander Löser – Beuth University of Applied Sciences Berlin
• Prof. Dr. Klemens Budde – Charité – University Medicine Berlin