How can mathematical modelling aid medical decision making?

Tuesday, June 15 at 04:15am (PDT)
Tuesday, June 15 at 12:15pm (BST)
Tuesday, June 15 08:15pm (KST)

SMB2021 SMB2021 Follow Monday (Tuesday) during the "MS06" time block.
Share this


Jasmina Panovska-Griffiths (University of Oxford), Eduard Campillo-Funollet (University of Sussex)


Mathematical and computational modelling can mimic real-life medical and biological environments to help us understand temporal and spatial behaviours and aid future predictions. The essence of modelling lies in designing sophisticated models that can help us replicate the biology and physiology we know, and allow us to make wise and reliable future decisions. In this mini-symposium we will showcase how mathematical modelling can be used to personalise radiotherapy, control the spread of the current pandemic, aid the process of deep tissue healing without scarring and improve non-invasive diagnosis of brain gliomas. Our speakers will take time to put the modelling in the context of real-life, discuss the theoretical framework of their models and outline the policy decision applications. The key of the modelling we will showcase is that it is currently being used by policy decision makers and for diagnostic and prognostic purposes. Therefore, the overarching aim of this mini-symposium is not only to showcase the specific models and the mathematics behind them, but also to discuss that creating a model requires far more than just raw data and technical skills; it also requires a close collaboration between mathematicians/computational scientists with clinical researchers and policy decision makers. Specifically, a good model requires robust theoretical framework, but also flexible multidisciplinary collaboration. The speakers will show how this is done in their research fields. Modern modelling approaches are a powerful tool for decision making, especially in this era of ”big data” and in the face of uncertainty. Therefore, in the course of this symposium we will showcase a range of topics revolving around medical and health applications of modern mathematical modelling, including cell biology, epidemiology and diagnostics, with the common denominator aiming to improve the medical decision making process and public health.

Elizabeth Ford

(Brighton and Sussex Medical School)
"Can modelling of primary care patient records enable detection of dementia earlier than the treating physician?"
Timely diagnosis of dementia is a policy priority in the United Kingdom (UK). However, recent research shows that a third to a half of patients with dementia do not have a diagnosis recorded in their primary care patient record, and for those that get a diagnosis, it takes over three years for the diagnosis to be made. We explored using modelling to automate early detection of dementia using data from electronic health records (EHRs). We investigated: a) how early a machine-learning model could accurately identify dementia before the physician; b) if models could be tuned for dementia subtype; and c) what the best clinical features were for achieving detection. Using EHRs from Clinical Practice Research Datalink in a case-control design, we selected patients aged >65y with a diagnosis of dementia recorded 2000-2012 (cases) and matched them 1:1 to controls, giving a total of 95k patients. We trained random forest classifiers, and evaluated models using Area Under the Receiver Operating Characteristic Curve (AUC). We examined models by year prior to diagnosis, dementia subtype, and the most important features contributing to classification. Classification of dementia cases and controls was poor 2-5 years prior to physician-recorded diagnosis but good in the year before. Features indicating increasing cognitive and physical frailty dominated models 2-5 years before diagnosis; in the final year, initiation of the dementia diagnostic pathway (memory loss symptoms, screening and referral) explained the sudden increase in accuracy. This leads us to think that automated detection of dementia earlier than the treating physician may be problematic using only primary care data, and that linking multiple sources of healthcare data could improve model performance.

Robin Thompson

(University of Warwick)
"Can modelling be used to predict whether or not the novel coronavirus will spread in the UK?"
The most devastating infectious disease epidemics are those that have a wide geographical range, as opposed to being confined to a small region. Early in the COVID-19 epidemic, an important question was whether or not SARS-CoV-2 would spread elsewhere and cause local outbreaks outside of China. A vital factor was the probability of establishment whenever a pathogen arrives in a new location, since this is a key component of any pathogens pandemic potential. We assessed this in real-time during the COVID- 19 epidemic. In this talk, we show how the probability of sustained transmission in other locations can be estimated from data that are available during infectious disease outbreaks. We show how estimates can be extended to include features such as transmission from paucisymptomatic infectors (infectious individuals with few symptoms). If time allows, we will also show how estimates can be generated for other pathogens and other epidemiological settings.

Fred Vermolen

(Delft University of Technology)
"Can modelling aid the process of deep tissue healing without scarring?"
Deep tissue injury is often followed by contraction of the scar. This contraction is caused by the pulling forces exerted by myofibroblasts and fibroblasts, which are cells that are responsible for the regeneration of collagen. In this talk, we will review several mechanical frameworks, such as viscoelasticity and morpho- elasticity, in which the latter framework can be used to simulate plastic deformations. Furthermore, we will consider cell-based as well as continuum simulation frameworks and some remarks about our upscaling efforts will be given. These upscaling strategies currently incorporate the relation between the use of the immerse boundary method and smoothed particle approach. Since many input parameters are patient-dependent, we will also present some results from the quantification of uncertainty that we have carried out.

Jasmina Panovska-Griffiths

(University of Oxford)
"Can combining modelling and brain radiomics non- invasively stratify brain gliomas?"
Combining MRI techniques with modelling is rapidly gaining attention as a promising method for staging of brain gliomas. This study assesses the diagnostic value of such a framework applied to stratify- ing treatment-nave gliomas from a multi-center patients into WHO grades II-IV and across their isocitrate dehydrogenase (IDH) mutation status. 333 patients from 6 tertiary centres, diagnosed histologically and molecularly with primary gliomas were retrospectively identified. Shape, intensity distribution and tex- ture features over the tumour mask were extracted. A random-forest algorithm was employed (2-fold cross-validation, 250 repeats) to predict grades or mutation status using the extracted features. Shape, distribution and texture features showed significant differences across mutation status. WHO grade II-III differentiation was mostly driven by shape features while texture and intensity feature were more relevant for the III-IV separation. Increased number of features became significant when differentiating grades further apart from one another. Gliomas were correctly stratified by mutation status in 71% and by grade in 53% of the cases (87% of the gliomas grades predicted with distance less than 1). Combining brain radiomics with modelling presents a promising approach for non-invasive glioma molecular subtyping and grading.

Hosted by SMB2021 Follow
Virtual conference of the Society for Mathematical Biology, 2021.