With improved care and treatment, more individuals are living longer after their heart attack. Unfortunately this means that the chance of developing other conditions over time after a heart attack has increased. Research to look at disease trajectories in these patients, to help prevent or interrupt fatal disease pathways is currently being carried out by LIDA’s Dr Marlous Hall, Associate Professor in Epidemiology. Her research group, made up of statisticians, data scientists and clinical research fellows, use advanced statistical and machine learning methods to understand the patterns of development of conditions over time in studies involving millions of electronic healthcare records.

Dr Hall has a particular focus on the potential impacts resulting from real-world health applications of data, with expertise in advanced analytical techniques applied to electronic healthcare data and the processing and analysing of large, complex clinical datasets. Her work on heart attack survivorship is funded by her Wellcome Trust personal fellowship as well as Alan Turing Institute and British Heart Foundation research grants and is aligned with her specific interest in multi-morbidity and long-term disease trajectories. “Most research provides a snapshot of multiple long term conditions at one point in time, but I’m interested in the life-course of a patient, looking at which conditions occur, in which order and how they progress over time,” says Dr Hall.

A key challenge is that many detailed clinical datasets are disease-specific. Clinical trials are vital in determining how effective treatments are for specific conditions, but tend to exclude, by design, complicated real-world patients, the majority of whom have multiple conditions. “By working with real-world electronic health records covering everyone in hospital across the entire country, we can be much more representative in terms of inclusivity, and understand real-life people with multiple diseases,” she explains.

In collaboration with colleagues from the University College London, University of Leicester, Karolinska Institute Sweden, University of Manchester, University of Oxford, Health Data Research UK, the Alan Turing Institute, as well as clinical academics at the University of Leeds, the team are analysing disease trajectories following a heart attack in 145 million anonymised hospitalisation records. Dr Hall explains: “The research is really about finding the best methods to get the best out of the health care data we have. Long-term trajectories of all possible conditions haven’t been explored in the past because of the complexities involved – and some of the methods required have not-yet been developed.”

Of particular focus is investigating the use of process mining techniques to understand disease pathways. Process mining is a deep-dive data analysis method utilized in business. Previously adopted in healthcare for treatment pathways, it has never-before been used in the long-term over a patient’s life-course, and here we use it to help understand general trends through machine learning. Examining hospital records using these machine learning techniques could help prevent or interrupt fatal disease pathways by revealing and better understanding the sequence and temporal pathways of many diseases, even discovering some that might otherwise have remained hidden..

“New treatments and interventions could be developed for patients with specific disease pathways to ultimately improve more lives,” Dr Hall explains. “We need to find out, though, what the best methods are for studying large amounts of data on thousands of unique disease codes in a sequence at unstructured time intervals. It’s why we are undertaking several research projects looking at the performance of process mining techniques and other machine learning methods when applied to vast quantities of complex, real-world patient data.

The research will provide a focus for where efforts should be directed to develop new ways of mapping complex disease pathways and trajectories. By looking at a population, the impact of disease progression over time and predicting what diseases a person might develop in future, it will be possible to identify the worst pathways, and intervene to lessen the severity for the patient, and lessen the burden on the NHS.

Find out more about Dr Hall

Dr Hall is engaged in a number of collaborative research programmes looking at multimorbidity and long-term trajectories. These include data phenotyping longitudinal multimorbidity trajectories in cardiovascular disease; examining healthcare utilisation and clinical outcomes among survivors of acute myocardial infarction; and, the Virtual Cardio-oncology Research Initiative (VICORI) aimed at joining together national health records to investigate relationships between cancer and cardiovascular disease. Dr Hall is also Deputy Head of the Clinical and Population Sciences Department in the Leeds Institute of Cardiovascular and Metabolic Medicine,  leads the Survivorship and Multimorbidity Epidemiology Research Group, supervises a number of PhD students, provides mentorship formally and informally to a range of academics and works hard to role model an open, inclusive and collegiate research culture.

Find out more about the Survivorship and Multimorbidity Epidemiology Research Group