Abstract

Over the last decades we have made major progress in diagnosing and treating cancer, however, this disease remains one of the leading causes of morbidity and mortality worldwide, responsible for millions of death each year. A 'moonshot to cure cancer' is being called for, which could translate this enormous progress into tangible outcomes for patients. Every patient is unique. No individual tumour has ever been observed before, or will ever be observed again, due to the enormous genetic/epigenetic heterogeneity between and within tumours and patients, causing each patient to react differently to drugs. To be able to provide 'the right drug at the right dose for every patient', we propose here to develop demonstrators, based on a deep molecular characterisation of tumour and patient as input to virtual patient models of the individual patient in silico. Overall, ITFoC will showcase federated activities on breast cancer to propose an advanced TRL demonstrator (TRL 5-6) in digital medicine together with a risk/benefit analysis of regulatory, ethical and economic issues, and a proposal of policy and guidelines.

Objectives

    ITFoC will focus on:
  1. The construction of one demonstrator for breast cancer to be tested in hospitals after 36 months (HEGP, Nantes), comparing i) European medical standards of care and ii) the use of probabilistic mechanistic and systemic models to guide clinical decisions.
  2. The proposition of a benchmark test of high quality anonymized curated data to compare the performance of the models in terms of sensitivity and specificity. In consultation with ethicists, a global challenge will be initiated to select the best simulation approaches to predict drug response.
  3. The development of a joint platform with access to different modelling methodologies and different business models addressing issues such as data standards, privacy, cybersecurity and data integration strategies.
  4. The evaluation of the benefit/risks, the ethical, regulatory, financial and societal impacts at the European level, acceptability and country-specific market access potential.
  5. The proposal of policy and guidelines for implementation of digital medicine at the European level.
  6. The dissemination of outcomes and analyses to key stakeholders including policy makers, patients groups, health professionals and the scientific community.

Estimated results

Machine learning based solutions as alternative to the mechanistic models. These are capable of providing the same output as the mechanistic models, but in a fraction of time, leading to order of magnitude improvements in terms of computational time and energy efficiency. Starting from the existing mechanistic computer models, which are mainly based on large systems of ordinary differential equations implemented on regular CPUs, and previous Graphics Processing Units (GPU) based solutions proposed for systems biology, new energy and time efficient, and parallel implementations of the models will be developed by employing multiple GPUs.