Uncertainty Quantification for Healthcare and Biological Systems

In recent years, complex computer models have been used more and more frequently for numerical modelling of biological systems underpinning complex diseases, with the goal of fully realizing the potential of personalised healthcare, and delivering tailored treatments for patients with debilitating health problems. Additionally, the increased understanding provided by in silico experimentation can reduce the need for animal experimentation. However, the potential impact of population-wide healthcare decisions means that incorporating complex computer models into the clinical settings must be done in a robust, transparent, and formalised way. As such, in silico assessment is starting to become part of the regulatory process, with guidelines such as those from the American Society of Mechanical Engineers (``Verification and Validation in Computational Modeling of Medical Devices'') being developed, where uncertainty quantification (UQ) is considered as an important part of the credibility assessment.

The field of uncertainty quantification (UQ) has grown around the idea that computer model analysis should formally take into account the various sources of uncertainty, namely code uncertainty, parameter uncertainty, model discrepancy and observation error involved in the system prior to performing model-based inference and decision support. Although engineering and physics models are well represented in the UQ field, the underlying assumptions differ significantly from healthcare and biological systems models: quantifying the uncertainty in healthcare models can pose different challenges.

Overall, the application of UQ to healthcare models is, relative to more traditional applications, still in its infancy. Methodological advances and greater uptake are needed to realise its full potential. The main objective of “Uncertainty Quantification for Healthcare and Biological Systems” workshop is to identify UQ challenges for mechanistic healthcare models by bringing applied mathematicians, statisticians and healthcare modellers together. In the long term, we also hope that the workshop can help develop a network of healthcare modellers and UQ experts, which will greatly improve the potential of healthcare models, aiding the reliability and reproducibility of model-based inference in healthcare.