Periodic Reporting for period 1 - STRATIF-AI (Continuous stratification for improved prevention, treatment, and rehabilitation of stroke patients using digital twins and AI)
Berichtszeitraum: 2023-05-01 bis 2024-10-31
Stroke is one of the most common causes of death, and a major contributor to serious, long-lasting handicaps. It is therefore of high importance to both prevent strokes from happening, and to treat and rehabilitate patients optimally, when a stroke has occurred. A complicating factor is that all phases of this journey involve different actors, data, and patho-physiological processes, which are currently not integrated with each other. We have in previous projects developed a world-unique technology, which could fundamentally change this situation: physiologically based digital twins. What makes our digital twins unique is that they describe physiological processes in all relevant organs, including their cross-talk, and including a connection to statistical machine-learning models. However, these digital twins have not yet been tested and implemented in stroke prevention and care. The objective of this project is therefore to make use of digital twins and AI, to aid in both prevention, acute treatment, and rehabilitation of stroke patients. The goal is that this AI-approach should lead to continuous stratification (hence the name: STRATIF-AI), meaning that any time new data about the patient is produced, these data update the patient's digital twin, which then leads to an update of the patient's diagnosis and treatment assessments. This is obtained by having all data about a patient be copied to that patients Personal Data Vault, which together with the twin and a visualization engine makes out the back-end of our new platform. This backend is then communicating with a series of different eHealth applications, which are used both by patients and their caregivers. In this project, we will both develop these twin models, the technology, and test it in practice, in two different clinical studies: one larger study on prevention (N=300), and a pilot study for rehabilitation. The potential of this project is to lay the basis for a new expandable healthcare system, which supports P4-medicine: Preventive, Predictive, Participatory, and Personalized medicine.
In these first 18 months of the project, we have made all necessary planning steps, which will underlie the rest of the work:
i) We have a data management plan, and searchable databases in all main data-providing partners
ii) We have a detailed requirement specification of the STRATIF-AI platform, which details both clinical UseCases, component specifications, an overall architecture, and an ethical inspection framework
iii) We have submitted ethical applications and startup packages for all 6 clinical studies: 4 of which will be used to collect data to train the models, and two that will test the new technology on real patients
iv) We have detailed communication and exploitation plans.
i) We have a data management plan, and searchable databases in all main data-providing partners
ii) We have a detailed requirement specification of the STRATIF-AI platform, which details both clinical UseCases, component specifications, an overall architecture, and an ethical inspection framework
iii) We have submitted ethical applications and startup packages for all 6 clinical studies: 4 of which will be used to collect data to train the models, and two that will test the new technology on real patients
iv) We have detailed communication and exploitation plans.
We have developed a prototype of our new platform and prevention app, which already is mature enough in certain parts to be used in commercial end-usage scenarios and demonstrations. The first release of the entire platform will happen in May 2025, and shortly thereafter, the first pilot study will commence. We have also published new validated versions of the underlying models. These show, among other things, that we can simulate the impact of lifestyle changes, such as diet, exercise, alcohol, and blood pressure medications, on both short-term and long-term physiological processes (e.g. metabolism, blood flow and pressure, weight, etc), as well as the resulting risk of a stroke. This is only possible because of our hybrid and multi-scale modelling approach, which combines all main organs in a mechanistic way, which in turn is connected with statistical risk models. Our already started exploitation work shows that companies seem to already now be willing to join co-development projects, also in other sectors than the ones we have initially targeted, such as company healthcare.