We are excited to announce that we have secured new funding from ALF, the state-funded research program for clinical research in collaboration with Swedish regions. The grant has been awarded to Gunnar Cedersund, the coordinator of STRATIF-AI, for a project titled “Implementation and Evaluation of a Digital Twin-Based eHealth Application for Preventive Health Dialogues.”
This project is directly linked to Clinical Study 2 within STRATIF-AI, which assesses our application designed for use during and after preventive health dialogues at primary healthcare centers. In this study, we will compare standard clinical practice with 300 health dialogues incorporating the new app.
With this additional funding, we can enhance app development and expand the clinical study to include 400 patients. The project will span three years, with an annual budget of 900,000 SEK, totaling approximately 3 million SEK (around 300,000 EUR). This support will significantly bolster our efforts in the field of preventive healthcare.
In this workshop, which will be held the 5th of November, we will discuss hybrid approaches, combining traditional bioinformatics and machine learning, with new generative AI approaches and mechanistically based digital twins. This combination of methods is ideal for dealing with a variety of different data types, and for developing AI models that are explainable and trustworthy. The event will feature keynote lectures, talks selected from abstracts, a poster session, and will be in hybrid format. Link to more info here:
A few weeks ago, on June 10, STRATIF-AI had its first meeting with fellow EU projects funded under the same call, all focusing on the use of AI for improved patient stratification. This initial two-hour meeting served as an introduction to each project and a discussion of key shared challenges (see Figure below). These challenges include data interoperability, federated learning, trustworthy and explainable AI (xAI), and MDR certifications.
During the meeting, we presented some core concepts of STRATIF-AI, such as our hybrid approach that combines multi-level, multi-timescale, and multi-organ simulations using mechanistic models as inputs to machine-learning models. This hybrid approach enhances the explainability of our models.
Furthermore, we explained shortly how STRATIF-AI uses data harmonization in two places: i) in connection to the federated learning set up by a collaboration between Catalina Martinez Costa (University of Murcia) and Lucia Gregorio Rodriguez (TREE), and ii) in connection to integrating a copy data from a variety of different sources into a personal data vault (Jesper Fellenius at Z2).
This collaboration and network of related projects will be useful for us to use in various related topics as we move forward, and we from STRATIF-AI are looking forward to new meetings. Thanks to the team at PREPARE for setting this meeting up.
We have a new part of the digital twin published! This time the new part, or sub-model, concerns the fat-driven disease etiology in the liver, based on fat fluxes in and out of the liver (Fig B below). As can be seen in Fig A, the model agrees with data for MRS PDFF, which is a magnetic resonance-based way of measuring liver fat, as well as with data for De novo lipogenesis (DNL), and e.g. ketone production and VLDL-TAG (Fig 2C-D in the paper). As usual, the model can describe data from different clinical studies, covering different diet and drug interventions in different cohorts, and the model can also successfully predict new data not used for training. This new sub-model is now being connected with the other meal and drug response models in the overall digital twin backend, and will thus become an integrated part of the future digital twin applications.
Our work “A physiologically-based digital twin for alcohol consumption—predicting real-life drinking responses and long-term plasma PEth“, has now been published in npj Digital Medicine.
In this manuscript we establish a mathematical model describing short-term alcohol dynamics, with or without simultaneously eating. We then go on to use this short-term model to make predictions about the appearance of the long-term alcohol marker Phosphatidylethanol (PEth). PEth has shown promise in monitoring alcohol consumption and by utilizing modelling the predictive ability of one’s drinking habits could be improved. Further, these predictive capabilities could make PEth-testing feasible in cases of liver diseases, such as MASLD.
You can read the article at: https://www.nature.com/articles/s41746-024-01089-6
On April 11-12, we had a new bi-yearly meeting in STRATIF-AI. This meeting was held in beautiful Nottwil, Switzerland, where the Swiss Paraplegic Research center (SPF) is situated. SPF is responsible for the work with policy and actions toward stakeholders to achieve ultimate clinical implementation, and they were also the hosts of this event.
We have these events every 6 months, and they serve both the purpose of team building and increased understanding of each others’ perspectives and serve to help us focus on the most urgent and timely topics. This time we focused primarily on the design of the 6 clinical studies, which ethical applications were to be submitted at the end of that month. Four of these studies are dedicated to the collection of patient data, aimed at training both the machine learning algorithms and the mechanistic aspects of digital twins. The remaining two studies focus on the real-world testing of eHealth apps within clinical settings. The largest of those will test whether the digital twin improves clinical health conversations, in 300 patients, compared with 300 matched controls.
Overall, the meeting was a success, and we have now passed the planning phase of the project, with requirement specifications and ethical plans, and are now moving into action: into prototype development of the apps, and towards the first pilot studies.