Work Packages

Goal

Gathering and preparing the patient data needed to build, train, and test the digital twins.

Purpose & Research Questions

• Bring together existing data from large international, national, and local patient databases and registries for Pulmonary Hypertension, alongside general population data (e.g., UK Biobank).
• Run a prospective clinical study to collect new, detailed data from PAH patients over time, using standard care methods, hospital based investigation and remote monitoring.
• This real-world data collection is vital for driving and evaluating the DT demonstrator.

Goal

Building the secure and efficient digital environment where patient data is stored, analysed, and linked together to form the network of digital twins.

Purpose & Research Questions

• Create the core technical framework and infrastructure, such as a secure research environment (SRE).
• Develop the 'digital thread' for each patient (containing all their data, simulations, forecasts) and weave these threads together into a 'digital tapestry'.
• Utilise relational databases, imaging databases, automated data analysis pipelines (for imaging, wearables), population graphs, and knowledge graphs to structure the data and represent relationships between patients, enabling the network functionality of WPs 3-5.
• Develop user interfaces for clinicians and potentially patients.

Goal

Developing the methods to create individual digital twins and ensure they accurately reflect each patient's unique cardiovascular physiology.

Purpose & Research Questions

• Implement a range of computational models (from simpler lumped parameter models to complex multi-physics heart models).
• Develop robust and efficient methods to calibrate these models (e.g. using AI, surrogate models, variational inference), adjusting parameters so outputs match specific patient data.
• Focus on making calibration scalable (handling many patients), fast (for timely updates), and able to leverage prior information from the network.

Goal

Developing the tools to use the calibrated digital twins to predict future health outcomes and test the potential impact of different scenarios or treatments.

Purpose & Research Questions

• Build on calibrated twins (from WP3) to provide sufficient and reliable predictive power.
• Develop and evaluate novel forecasting techniques (e.g., NARX models, Transformer-based approaches, mixed-effects models) to predict clinical worsening events and changes in clinical measurements over 6-12 months.
• Enable 'what-if?' testing to simulate patient responses to different treatments or physiological changes.

Goal

Developing and applying methods to leverage the entire network of digital twins to improve the performance and efficiency of individual twins.

Purpose & Research Questions

• Explore the unique benefits of an interconnected population of DTs.
• Test methods like Transfer Learning, Multi-task Learning, Ensembles, developing similarity metrics, and using geometric data transforms.
• Reduce the cost and improve the accuracy of calibrating and forecasting DTs by exploiting collective information within the network.

Goal

Ensuring the digital twin technology is ethically developed, addresses real needs, is ready for adoption, and engages effectively with all stakeholders.

Purpose & Research Questions

• Evaluate and assure the DT demonstrator using frameworks like NASSS and the TEA Platform to identify barriers and facilitators to adoption.
• Conduct ongoing engagement with patients and clinicians through focus groups and co-design sessions.
• Actively map and address ethical considerations (data ownership, algorithmic bias, transparency) through systematic research.
• Manage public communication, patient engagement, and community building, ensuring integration with other work packages.