THROMB-AI: Computational risk stratification of thrombus formation and lysis in post-stroke patients

Description

Cardiovascular disease is the leading cause of mortality and morbidity in developed countries. Often, the main pathological process responsible is the development of an arterial thrombus (or blood clot). The resultant blockage of arteries leads to drops in blood flow and eventual tissue death, such as in ischaemic stroke. There is a natural balance in the blood between agents that stimulate thrombus formation, and agents that break down clots (also known as thrombolysis). In disease, this balance can be disrupted leading to thrombus formation. As such, the usual treatment for such conditions is the use of anti-thrombotics to reduce the propensity for further thrombus formation. However, with an increasingly complex, multi-morbid and ageing population, balancing thrombotic and bleeding risks becomes more difficult. Therefore, being able to personalise risk and tailor treatment will aid clinicians in making right decisions and improving patient outcomes.

The Global Thrombosis Test (GTT) is a point-of-care test that measures a patient’s blood clotting capability and thrombolytic activity1,2. This project will use the GTT outputs of time-to-occlusion and time-to-lysis, along with patient data and mathematical models to predict patients at risk of arterial thrombotic events and their relative risk of anti-thrombotic use.

The objectives of the PhD will be split into 3 work packages:

WP1: Data collection - Sequential GTT in a post-stroke cohort

As part of an ongoing study at the University of Liverpool, the PhD candidate will analyse bloods taken from post-stroke patients using the GTT, obtaining personalised curves of time-to-occlusion and time-to-lysis for each patient. Patients will be followed up after 3 months for a second blood draw. Clinical outcomes assessing adverse cardiovascular events will be collected at 12 months.

WP2: Mathematical modelling - Personalised in silico model of thrombus formation and lysis

Using the data gathered, the candidate will build upon prior mathematical models of thrombus formation that are dependent on shear strain rate3. The parameters used to model thrombus formation and lysis will be fitted to patient data collected in WP1 – effectively developing personalised models of thrombus formation, lysis, and the impact of anti-thrombotics on an individual. These models will be used, in combination with blood flow models, to determine whether certain patients are at risk of heart attack and stroke in specific arterial regions (e.g. coronary arteries).

WP3: Machine learning - Clinical risk prediction

This WP will combine outcomes from WP1 and WP2 to predict clinical outcomes such as death and ischaemic events. The risk prediction model will be built using machine learning methods e.g. random forests and support vector machines. We will compare if the addition of GTT times and information from the personalised in silico models significantly improves risk prediction beyond that possible with standard clinical parameters4,5.

The work to be undertaken will be conducted at the Liverpool Centre for Cardiovascular Science as a collaboration between biomedical engineers (Dr El-BouriDr Narracott) and clinical experts in cardiovascular disease and thrombosis (Prof LipDr GueProf Gorog). Training will be provided in using the GTT. We are looking for a highly motivated student with a strong background in engineering/mathematics/computer science or similar background.

Enquiries to: Dr Wahbi El-Bouri ()

To apply: Applicants should send a CV and a covering letter to Dr Wahbi El-Bouri, 

Availability

Open to students worldwide

Funding information

Self-funded project

Supervisors

References

1. Yamamoto, J. et al. Görög Thrombosis Test: a global in-vitro test of platelet function and thrombolysis. Blood Coagul. fibrinolysis an Int. J. Haemost. Thromb. 14, 31–39 (2003).
2. Otsui, K. et al. Global Thrombosis Test – a possible monitoring system for the effects and safety of dabigatran. Thromb. J. 13, 39 (2015).
3. Mehrabadi, M., Casa, L. D. C., Aidun, C. K. & Ku, D. N. A Predictive Model of High Shear Thrombus Growth. Ann. Biomed. Eng. 44, 2339–2350 (2016).
4. Sharma, S. et al. Impaired thrombolysis: a novel cardiovascular risk factor in end-stage renal disease. Eur. Heart J. 34, 354–363 (2013).
5. Taomoto, K. et al. Platelet function and spontaneous thrombolytic activity of patients with cerebral infarction assessed by the global thrombosis test. Pathophysiol. Haemost. Thromb. 37, 43–48 (2010).