>
Original Press Relase [German]
When every hour counts: Sepsis early detection already possible with basic blood tests - thanks to new AI-based methods
AMPEL project at the University of Leipzig Medical Center achieves milestone in patient safety / Previous sepsis parameter significantly surpassed / AMPEL as digital infrastructure for clinical AI applications in routine care
Dr. Daniel Steinbach, Martin Federbusch und Maria Schmidt (f.l.t.r.) are part of the AMPEL core team. They are further developing AMPEL into a
digital infrastructure that enables clinical AI applications in routine
care. Their study on predicting sepsis based on the complete blood count
has now been published in the world's leading journal for laboratory
medicine, "Clinical Chemistry.
Leipzig.
With new methods based on artificial intelligence (AI), the AMPEL project at University of Leipzig Medical Center achieves a significant advancement in patient safety: Now, basic blood tests are sufficient to detect sepsis in patients earlier than before.
Even the well-known sepsis parameter Procalcitonin can be significantly surpassed.
Since its inception in 2018, AMPEL has continued to evolve and is now a comprehensive digital infrastructure within the clinic, enabling clinical AI applications in routine care.
Over the past five years,
the multi-award-winning project has been providing crucial support to nursing and medical staff in patient care. By detecting critical situations in real time, it has significantly enhanced patient safety. Automated alerts improve the availability and weighting of medical information. AMPEL utilizes simple calculations to complex AI models and monitors all necessary data live.
When there is suspicion of sepsis, it can be critical, and every hour counts. The survival of this often fatal disease depends significantly on the earliest possible administration of antibiotics. The AMPEL project at the University of Leipzig Medical Center has now achieved a milestone in sepsis early detection: With new methods of machine learning, the team was able to develop an AI model and scientifically confirm it at two additional sites, in Germany and the USA. "Our study on predicting sepsis based on the complete blood count has been accepted by the world's leading journal in laboratory medicine, 'Clinical Chemistry'. We showcase the potential of utilizing AI methods alongside a minimal set of pre-existing routine data to establish a continuous screening process for patients with early signs of sepsis", explains Dr. Daniel Steinbach, physician and research associate at the
Institute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics and at the
Data Integration Center of the University of Leipzig Medical Center as well as a member of the AMPEL core team. "And the fact that, without additional costs, our AI model significantly outperforms the prediction of the established marker Procalcitonin (PCT) is likely to generate substantial interest."
Labor values from blood count tests are always available but seldom utilized
The idea that the data from few parameters of the complete blood count could help detect sepsis early came as a surprise even to the interdisciplinary AMPEL team. "The initial results showed that the developed AI models often rely on the data from these basic blood tests. These are laboratory values that are always present but are rarely considered," says Dr. Steinbach. In the clinical management of sepsis detection, these laboratory findings hardly play a role, according to the expert, although they are determined in almost every hospital with almost every laboratory test. Instead, emphasis is placed on specific laboratory parameters such as Procalcitonin. Procalcitonin is contradictory in itself, says Maria Schmidt, who is also part of the AMPEL core team: "Although it is used almost everywhere, studies regularly conclude that its predictive power is too low. What remains is a recommendation for guiding the appropriate antibiotic therapy during the course of the disease." In their sepsis study, they have now shown that with machine learning methods and data of the complete blood count, the predictive power of Procalcitonin can be significantly improved, explains biometrician Schmidt. After the theory, practical application is now to follow: "In laboratory medicine, I am not aware of any AI model that has been tested as extensively and comprehensively as our published sepsis model. Nevertheless, in the end, it remains only theory, and only practical application will show whether and how much support it really provides. Fortunately, that is precisely one of the core competencies of AMPEL", says Dr. Steinbach.
Further development as an Open Source project
Martin Federbusch, a specialist in laboratory medicine, leads the AMPEL project and regrets that the University of Leipzig Medical Center is still the only hospital in Germany equipped with a digital infrastructure comparable to AMPEL. "Thus, an important goal for us remains: the expansion to additional locations", emphasizes Federbusch. "Thanks to the extensive support at our location, we are able to independently advance the AI platform AMPEL." Since this year, AMPEL has been further developed as an Open Source project under the leadership of University of Leipzig Medical Center by Prof. Toralf Kirsten and his Department of Medical Data Science. The focus is on creating a non-profit AI infrastructure for healthcare that meets the highest standards of adaptability, interoperability, and transparency.
> Original Press Relase [German]
> More information about the sepsis model
