Why don’t patients take their pills? It’s an issue that has plagued healthcare providers and clinicians for years. Chronic diseases are the leading cause of death in New Zealand and yet many patients are not receiving the benefits of care because they are failing to take their medication on time and/or in the correct dose.
A psychological model that attempts to explain and predict health behaviours has been created called the Health Belief Model (HBM). Previously, data for this model has been gathered using questionnaire surveys, but this has proved too limiting. Postgraduate doctoral student Feiyu Hu and his supervisor Professor Jim Warren from the University of Auckland are looking at how Big Data analytics can be used to determine if patients’ geographical information, combined with their medication history, will provide better insight into this perplexing problem.
The research, entitled Spatio-Temporal Big Data Analysis of Adherence Behaviour in Chronic Disease, will initially focus on cardiovascular disease; however, it is expected that the findings can be applied to other chronic diseases.
Data will be linked from various datasets, including the national health index (gender, ethnicity, age), pharmaceutical claims, hospital diagnoses, mortality records and New Zealand census data. The data will be analysed and compared using a range of methods and applications, to confirm to what extent spatial and time series models can improve the prediction of medication adherence.
In the second stage of the research the project will broaden to include a wider data sample, and in addition to cardiovascular disease, diabetes will also be studied (the two diseases are often interlinked). The spatio-temporal models developed and tested in the initial stage will then be applied to this richer data set. The researchers are expecting that the addition of information relating to employment history and education will reduce the significance of geographic data in determining better adherence models.
The study aims to produce the best-ever characterisation of social inequities as related to ethnicity, low income and poor education. It will therefore provide clinicians, policymakers and researchers with a valuable resource to use when looking for ways to achieve better medical adherence across the entire population. This will in turn make it easier to highlight those patients potentially at risk and ensure precious health resources, such as follow-up care by nurses, are deployed where they are most needed.
The models created by this study can also be applied to health behaviours such as diet, exercise and cessation of smoking. As data collection is set to grow exponentially with the internet of things, the ability to collate, analyse and provide meaningful information back to clinicians – as well as patients, and their families – will become increasingly useful.