The masses of patient data collected by different departments of Townsville Hospital contain many hidden answers to medical and health questions.
The answers, however, stay hidden unless the data can be linked up, organised and interrogated, which is exactly what is being done by a team of researchers, data scientists and clinicians from AITHM, Townsville Hospital and JCU’s College of Medicine and Dentistry.
Professor Damon Eisen of JCU’s College of Medicine and Dentistry said the team was focussing its attention on data for patients admitted to Townsville Hospital with infectious disease codes over an 11-year period between 2006 and 2016.
Data sources included the Queensland Admitted Patient Data Collection, Emergency Department Information Services, Pathology Queensland, pharmacy dispensing data, notifiable conditions and the National Death Registry. Once these data sources have a linkage code, they are anonymised.
The cohort includes more than 41,000 patients with nearly 380,000 admissions, around 2 million pathology results, and more than 1.8 million diagnostic/procedure codes.
“All of those pieces of information come together to provide a rich and detailed picture of individual patients and groups of patients. We can use it to consider a whole range of conditions.” said Professor Eisen.
Some questions the team have begun asking of the data relate to mortality in different demographic groups. They have also linked additional data from the Bureau of Meteorology with that of pneumonia patients to examine the impact of climate on pneumonia incidence in the Northern Queensland region.
“The organisation of the data is challenging. We have had a very brilliant data scientist finalising a relational model so that clinicians can readily use this resource, giving them an interface to get a read-out on questions they might have can make the data come alive,” said Professor Eisen.
The team working on the data linkage project includes AITHM Professor Emma McBryde, AITHM Research Fellow Dr Oyelola Adegboye, data scientist Matthew Murray and clinicians from Townsville Hospital.
Professor McBryde said other plans for the data included auditing the number of people with rare diseases (e.g. melioidosis) or severe outcomes (e.g. acidosis/ICU admissions) and examining risk factors for these.
“We are also very interested in the impact of being rural/remote on outcomes and the impact of resistance on outcomes. We plan to interrogate the database using machine learning algorithms to see if this provides any additional insights into predictors of death,” said Professor McBryde.
Machine learning approaches will allow the team to use the data that has already been collected to “train the machines” to identify and predict problems and patterns of diagnosis that are occurring in real time.
