PI: Gunnar Cedersund (LiU)
In 2019, the Swedish translation for “Digital twins” entered the public Swedish dictionary, SAOL. Digital twins are computer models, which can describe, e.g., the specific physiology in a patient. Nevertheless, all others’ digital twins only describe a single organ or function in the human body, such as blood flow. We have developed the only digital twin that combines such diverse aspects as brain activity, blood flow, metabolism, and inflammation into a single model. This is the result of almost 20 years of ongoing mechanistic modelling. A first prototype for this twin was launched at Almedalen last summer, and this has been followed by high-profile keynote presentations at the biggest medical institute in the world, the National Institute for Health (NIH), and by keynotes at national conferences such as “IT i vården”-dagen, ModProd, etc. However, to make this initial prototype usable in actual healthcare, there are remaining challenges that must be overcome. In this project, we will accomplish this in three steps. Step 1: to overcome the practical, legal, and ethical challenges with creating an information platform where patient-specific data and information from all relevant sources can be stored. This is made possible by Cory Robinson (CR, LiU), who is an expert on such ethical and legal issues, and by Erik Sundvall (ES, LiU), who is information architect at Region Östergötland, and responsible for handling electronic (EHR) and personal healthcare records (PHR). Step 2: to create new hybrid AI methodologies, which combines the strengths of mechanistic modelling and machine learning, and which allows e.g. for usage of a large variety of small- and large-scale data, to create personalized digital twins. This is made possible by the main applicant Gunnar Cedersund (GC, LiU), who combines his unique models with expertise from the EU-network Precise4Q (P4Q, precise4q.eu), where expertise on machine learning resides. Step 3: To design new Intelligent Assistants, which combines the digital twins with the new patient-specific data storages, and which is helpful for both medical doctors, patients and ordinary citizen. This is made possible because we have started a spin-off company, SUND, and joined forces with the Visual Sweden MeDigiT project. The strength of this project is that we developed ground-breaking hybrid AI approaches (Topic A) to be able to extract information from a wide variety of Big Data sources (Topic 2); that we also deal with legal, ethical and commercial challenges (Topic B); and that we combine all of this into an Intelligent Assistant tool (Topic 5). Because of this unique combination, this project will result in usable digital twins for healthcare. The twins will be tested at LiU (GC, ES, CR), in Lund (Karin Stenkula, KS) and in Germany and Spain (P4Q).
Project number: A12