Periodic Reporting for period 1 - SMASH-HCM (Stratification, Management, and Guidance of Hypertrophic Cardiomyopathy Patients using Hybrid Digital Twin Solutions)
Período documentado: 2024-01-01 hasta 2025-06-30
Recognized as a strategic innovation trend, Digital Twin (DT) technology has successfully been applied across various sectors and are now also emerging in healthcare, e.g. in the context of the Virtual Human Twin EU framework. SMASH-HCM explores the potential of DTs to enhance personalized healthcare, focusing on hypertrophic cardiomyopathy (HCM). HCM is the most common inherited cardiac disorder, leading to variety of symptoms, and potentially sudden cardiac death. The goal of SMASH-HCM is to develop a DT, or virtual model, of the heart and vascular system with sympathetic nerve control from cell, tissue, organ and finally system level thus integrating multi-scale and multi-organ spatiotemporal biophysical data from a multitude of sources. These models and their data are integrated with data-driven models utilizing clinical and unstructured data to form a SMASH-HCM decision support solution for healthcare professionals and patient empowerment.
Thanks to the EU HOP-ON program, SMASH-HCM was expanded to include molecular modelling and data-driven tools to screen transcriptomic and drug molecular data to find novel candidates for HCM through drug repurposing.
SMASH-HCM is formed by 8 research partners, 3 hospitals, 4 SMEs, and a global health-technology corporation, in collaboration with patients through patient organizations in four countries. The project will develop a DT platform to dramatically improve HCM stratification by integrating multilevel and multiorgan dynamic biophysical and data-driven models into a three-level deep phenotyping approach. SMASH-HCM advances the state of the art in human DTs by developing in-vitro tools, in-silico models from molecular to systemic levels, structured and unstructured data analysis, and explainable artificial intelligence - all integrated into a decision support solution for both healthcare professionals and patients.
Thanks to the EU HOP-ON program, SMASH-HCM was expanded to include molecular modelling and data-driven tools to screen transcriptomic and drug molecular data to find novel candidates for HCM through drug repurposing.
SMASH-HCM is formed by 8 research partners, 3 hospitals, 4 SMEs, and a global health-technology corporation, in collaboration with patients through patient organizations in four countries. The project will develop a DT platform to dramatically improve HCM stratification by integrating multilevel and multiorgan dynamic biophysical and data-driven models into a three-level deep phenotyping approach. SMASH-HCM advances the state of the art in human DTs by developing in-vitro tools, in-silico models from molecular to systemic levels, structured and unstructured data analysis, and explainable artificial intelligence - all integrated into a decision support solution for both healthcare professionals and patients.
A SMASH_HCM database, consolidating data from our clinical partners that later will be expanded with data also from in-vitro experiments and in-silico individual and population level synthetic data has been established and is accessible within the entire consortium.
We have made progress towards better understanding of the HCM mechanisms by developing in-vitro models including HCM patient cell lines and in-vitro cardiac tissue models as well as novel protocols to challenge the in-vitro cultures mimicking exercise environments. SMASH-HCM has developed multilevel cardiac HCM models from single cell to tissue and organ level including electrophysiology, mechanics and energetics. Similarly in cardiovascular models, we have developed multilevel simulation from smooth muscle cell electrophysiology and mechanics to vessel wall and systemic level.
We have developed first data-based models describing HCM clinical phenotypes using both clinical data as well as unstructured data mining methods.
Initial versions of requirements specifications for a decision support solution include identification of clinical needs elicited from HCM specialists and the needs of patients to gain information on their disease. This was turned into functional requirements for the platform. Additionally, we performed a study on the socioeconomic issues related to HCM. We developed an API and software architecture to effectively combine mechanistic and data-driven models developed by the different partners in a flexible manner, thereby laying the foundation for the digital twin implementation from a practical usage perspective.
We have made progress towards better understanding of the HCM mechanisms by developing in-vitro models including HCM patient cell lines and in-vitro cardiac tissue models as well as novel protocols to challenge the in-vitro cultures mimicking exercise environments. SMASH-HCM has developed multilevel cardiac HCM models from single cell to tissue and organ level including electrophysiology, mechanics and energetics. Similarly in cardiovascular models, we have developed multilevel simulation from smooth muscle cell electrophysiology and mechanics to vessel wall and systemic level.
We have developed first data-based models describing HCM clinical phenotypes using both clinical data as well as unstructured data mining methods.
Initial versions of requirements specifications for a decision support solution include identification of clinical needs elicited from HCM specialists and the needs of patients to gain information on their disease. This was turned into functional requirements for the platform. Additionally, we performed a study on the socioeconomic issues related to HCM. We developed an API and software architecture to effectively combine mechanistic and data-driven models developed by the different partners in a flexible manner, thereby laying the foundation for the digital twin implementation from a practical usage perspective.
We have developed an open-source tool for generating populations of in silico cardiomyocyte models, transitioning from bespoke research software to a broadly accessible platform. The new tool offers expanded functionality, cross-platform compatibility, scalable performance, intuitive design, and thorough documentation. Open release fosters transparency, community-driven development, and broader adoption. Feedback from users will further refine documentation and usability.
Our cross-phenotype in-silico cardiomyocyte translator extends existing literature by integrating active tension biomarkers via electromechanically coupled single-cell models and demonstrating that minimal experimental protocols can reliably predict key electrophysiological and mechanical outputs.
A novel in-silico model of the MYBPC3 c.772G>A mutation has been integrated into three adult ventricular cell models, revealing model-dependent variability in drug response. Simulations with dapagliflozin suggest additional mechanisms may contribute to action potential shortening, emphasizing the value of in-silico testing to guide experimental research.
We have also advanced repolarisation inference by enabling reconstruction of both normal and abnormal gradients directly from the ECG, validated against ground truth data. This provides a robust framework for mapping activation and repolarisation throughout the myocardium - including the diseased septum, not just the epicardial surface. This is expected to enable patient-specific digital twin models for HCM, offering insights into ECG abnormalities and supporting personalised arrhythmic risk assessment.
We have developed one of the first integrated models of the cardiovascular system from smooth muscle cell electrophysiology and biomechanics to vessel and finally system level responses, providing us new means to understand the complex interaction on vascular system regulations and the changes in vascular responses in HCM patients.
Our new unstructured data analytics have produced tens of novel models describing possible new clinical biomarkers for HCM stratification to be tested on SMASH–HCM database and on the biophysical models developed in WP4 and 5.
We have developed a software architecture and algorithm logic that allows to have mechanistic and data-driven models combined in order to address the challenging decision support problems related to HCM. It enables flexible interaction between models developed in different labs using different computing environments and languages.
Our cross-phenotype in-silico cardiomyocyte translator extends existing literature by integrating active tension biomarkers via electromechanically coupled single-cell models and demonstrating that minimal experimental protocols can reliably predict key electrophysiological and mechanical outputs.
A novel in-silico model of the MYBPC3 c.772G>A mutation has been integrated into three adult ventricular cell models, revealing model-dependent variability in drug response. Simulations with dapagliflozin suggest additional mechanisms may contribute to action potential shortening, emphasizing the value of in-silico testing to guide experimental research.
We have also advanced repolarisation inference by enabling reconstruction of both normal and abnormal gradients directly from the ECG, validated against ground truth data. This provides a robust framework for mapping activation and repolarisation throughout the myocardium - including the diseased septum, not just the epicardial surface. This is expected to enable patient-specific digital twin models for HCM, offering insights into ECG abnormalities and supporting personalised arrhythmic risk assessment.
We have developed one of the first integrated models of the cardiovascular system from smooth muscle cell electrophysiology and biomechanics to vessel and finally system level responses, providing us new means to understand the complex interaction on vascular system regulations and the changes in vascular responses in HCM patients.
Our new unstructured data analytics have produced tens of novel models describing possible new clinical biomarkers for HCM stratification to be tested on SMASH–HCM database and on the biophysical models developed in WP4 and 5.
We have developed a software architecture and algorithm logic that allows to have mechanistic and data-driven models combined in order to address the challenging decision support problems related to HCM. It enables flexible interaction between models developed in different labs using different computing environments and languages.