Throughout his 25-year career, Link枚ping University researcher Gunnar Cedersund has worked toward a single overarching goal: to create a digital twin of the human body and to use this twin as the foundation for transforming our healthcare system.
Methodological breakthroughs prior to M4-health
Between 2000 and 2018, Gunnar Cedersund laid the methodological foundation for, among other things, how sub-models of organs and cells can be systematically tested and refined using data and expert knowledge [references 1鈥�2, see below].
Other significant early methodological developments concerned how these sub-models could be combined into a coherent model of the body as a whole [3鈥�5]. These methods have been applied to the development of specific models for major organs, including the liver [6], pancreas [7], adipose tissue [8], brain [9], and heart [10]. Already from the outset, these methodological developments were funded by the Swedish Research Council (VR) through various projects, which in 2018 evolved into a series of projects linked to the concept M4-health.
Project 1:
M4-health: the foundation for general AI in healthcare
The concept "M4-health" has a double meaning. The name 鈥� standing for "Model for health"鈥� signifies the need for a new model of health, i.e. a new way of organising our healthcare systems. The name also has a more technical meaning, describing the type of digital twin models developed by Gunnar Cedersund: multi-level (ranging from intracellular to whole-body levels, including all major organs), multi-timescale (milliseconds to years), multi-species (humans, laboratory animals, and microphysiological systems), and mechanistic (describing actual biochemical and physiological processes). This concept distinguishes Cedersund鈥檚 modelling from others, both nationally and internationally, as other initiatives primarily focus on modelling of individual organs rather than how they can be combined into a model of the body as a whole.
The VR project conducted between 2019 and 2023 established the scientific foundation for bringing these M4 models closer to clinical application. A key aspect of this work involved not only further developing the mechanistic models but also devising methods for hybrid models that incorporate machine learning [11鈥�12]. This enables the integration of data from large-scale databases (used to train machine learning models via federated learning) and data from mechanistic studies of humans, organs, and cells into a single hybrid model. The mechanistic basis of these hybrid digital twin models also allows for their reuse across a wide range of applications 鈥� for instance, in the study of various cardiometabolic diseases, such as diabetes, hypertension, heart attack and stroke.
Project 2:
M4-health: Digital twins for the prevention and treatment of cardiometabolic diseases
In this follow-up VR project, the scope of the work has expanded to include a focus on end-use within healthcare. For instance, we have developed models for the characterization, prognosis, and diagnosis of specific conditions, such as liver metabolism and liver disease [13鈥�14], alcohol metabolism [15鈥�16], type 2 diabetes, and hypertension [17鈥�18].
This work has also involved the development of e-health applications that allow end-users to keep their digital twin on their mobile phones and exchange data, simulation results, and realistic animations, utilizing a central server where all twins and clinical data are managed and hosted. Furthermore, this initiative has led to commercialisation and the founding of several spin-off companies. Finally, the project has entailed leading various national and international networks that have advanced these digital twin models toward clinical trials and commercial end-use.
Parallel expansions into other major network projects
One of the initial network projects was the Vinnova-funded initiative "Digital Twins in Healthcare: Model Validation, Co-design, and Clinical Evaluation" (2021鈥�2023). Conducted in collaboration with Mattias Ekstedt (Region 脰sterg枚tland) and Karin Stenkula (Lund University), this project laid the groundwork for testing these twins in preventive health conversations aimed at preventing cardiometabolic diseases.
These health consultation initiatives are now also funded by a three-year ALF project titled "Implementation and Evaluation of a Digital Twin-based eHealth Application for Preventive Health Consultations" (2025鈥�2027). Furthermore, these consultations form a central component of the major EU project STRATIF-AI (2023鈥�2027; budget: SEK 6M Euro), which is developing and testing digital twins for stroke treatment and rehabilitation.
Finally, these digital twins have recently served as the foundation for the extensive Swedish network "M4-HEALTH: A New Model for Health via Digital Twins and AI." This network advances the application of these twins beyond healthcare to sectors such as the food industry, defense, education, and elderly care.
M4-HEALTH involves 16 Swedish universities and university colleges, engages over 100 companies and other organisations, and represents one of the most ambitious and comprehensive Swedish initiatives to date regarding the future of healthcare. All these network projects are coordinated by Gunnar Cedersund and utilize his reusable digital twins, which were initiated in the two VR-funded projects mentioned above.
