Zachary Waller (He/Him)

 

Current research project

Combining machine learning and casual inference to predict cardiovascular disease for use in a decision model

Cardiovascular disease remains the highest cause of death worldwide. Despite numerous measures of cardiovascular risk, a large number of patients go undetected based on classical risk factors alone, e.g. hypertension, smoking status and BMI. Novel biomarkers have potential for greater predictive accuracy, but choosing an optimum panel of biomarkers remains a difficult task as some may merely be proxies for known risk factors, therefore providing little improvement in prognosis while being more resource intensive to measure. 

By utilising improvements in machine learning techniques a panel of biomarkers best suited to predicting cardiovascular events can be selected. Causal inference methods can further highlight those that are not only predictive but have a causal effect on the outcome. 

Multi-state Markov models can be used to model the transition of at risk individuals through health states based on traditional and biomarker risk factors, with cost-effectiveness analysis taking into account treatment and biomarker measurement costs to understand whether prevention strategies are both effective and cost-effective.

Biography

I graduated with a physics MPhys in 2015 from the University of Manchester. Since then I have worked mainly as a data scientist for the civil service including contributing to the development of a weekly excess mortality model for Public Health England. I started my PhD in October 2021.

Research interests

• Decision Modelling

• Cost-effectiveness

• Multistate models

• Causal Inference

• Machine learning