Map-based Visualisation and Statistical Inference with Dynamic Health Data

Whenever someone visits their GP or the hospital, their medical condition (incl. test results, vaccinations, etc.) is recorded in their personal NHS health file. This process generates a largely untapped wealth of data on the nation's health. A health atlas uses this medical information to generate maps in order to show which areas of a region (eg. town or country) are at a low or high risk of a particular disease. When assessing the risk of disease it is common to account for factors such as poverty, age and coexisting diseases. Such health atlases already exist, but are inflexible as they use old historical data and take a long time to produce. Consequently they do not display the present health landscape.

The fundamental aim of the project is to develop methods and tools for creating 'real-time' health atlases; as an example it will use the NHS medical records of the population living in the city of Salford, UK (over 200,000 people). The reason for targeting Salford is that it has over twenty years of records stored in a central state-of-the-art computer system, everyone's NHS records are regularly collected therefore they are completely up-to-date (over 200 million events have been recorded). This therefore opens the opportunity to create the first 'real-time' health atlas which will display, partly through maps, the current health state of the Salford population in 'real-time' via user-friendly interactive web applications. There will be two versions of the atlas one for the general public as described above. The other version will convey more detailed/specialist information so will only be available to health care professionals and scientists. In both cases great care will be taken to ensure the information displayed is sufficiently straightforward to understand. Privacy will be strictly protected; individuals will not be identifiable on this atlas.

Within the lifetime of the project, the real-time health atlas will focus on kidney disease which affects approximately 5% of the adult population in the UK although it is far more prevalent in some sectors eg. in the elderly. Furthermore, people who suffer from other medical conditions, such as cardiovascular disease, high blood pressure and diabetes, have a higher risk of developing kidney disease. Certain ethnic groups are more likely to have these medical conditions, consequently are more likely to develop kidney disease, hence this disease is an important public health concern. Therefore mapping this disease's distribution on a real-time health atlas is a timely development.

The real-time health atlas is to be developed by a Research Fellow based in the 'Combining Health Information, Computation and Statistics' Group in the School of Medicine at Lancaster University. The Fellow will work with colleagues at the Farr Institute of Health Informatics Research, based within The University of Manchester, and also with the tertiary care group for kidney disease within the Salford Royal NHS Foundation Trust. These three institutions, each with strong research records in their own fields, will lend their expertise to advise on the statistical and software development behind the real-time health atlas. This atlas sits at the interface between statistics and information technology where the funder, The Medical Research Council, recognises there is a need to invest, partly due to a UK skills shortage.

This real-time health atlas will be designed so that in the future, beyond the lifetime of this project, it can be used both for other diseases and in different geographical regions. The atlas will be up to date and interactively display disease specific information on the internet therefore everyone, from individuals to health care professionals to policy makers, will have the opportunity to improve people's quality of life by targeting specific areas where there is a greater likelihood of certain diseases; in general this should reduce long-term health care costs.

The primary aim of the research is to develop the statistical and informatics methodology required for real-time health atlases; as an exemplar it will use primary care electronic health records (SIR) from the population of Salford city and focus on chronic kidney disease (CKD). This will extend the current static health atlases to real-time web applications which will be available in two forms; a 'public' web application will be aimed at the general public and a more sophisticated 'professional' web application will be made available to professionals (subject to data owner's permission). To achieve this statistical methodology will be developed along with open-source software for: querying SIR (SQL); statistical modelling (R-package); and interactively conveying the results via web applications (JavaScript) adhering to established human-computer interaction principles. The software will be flexible enough for the atlas, beyond the lifetime of the project, to be used for other diseases and in different locations. Furthermore it may inform future research projects or health care authorities if unexpected (spatial and/or temporal) disease patterns emerge.

The secondary aim is to train the Fellow so that she can make a career transition from physics to biostatistics and informatics. Formal statistics training will realised by her undertaking an MSc in statistics with a bias towards medical statistics and training in informatics will be via directed reading and short courses at the Farr Institute. The MSc project will give the Fellow additional experience of applying advanced statistical methods in a clinical setting; aim is to identify key risk-factors underlying chronic kidney disease progression to end-stage (outcome). Existing statistical methodology will suffice for modelling single time-to-event outcomes but joint modelling of multivariate time-to-event outcomes will require some novel methodological extension.

For the benefit of academic researchers and health care professionals wherever possible all research will be disseminated to the appropriate communities (nephrology, biostatistics, health informatics) by open-access publications and open-source software. The beneficiaries for each activity are as follows:

i) Research Activity A: real-time health atlas

The atlas will be realised by developing new methods and software including two interactive web applications for conveying the health information to either the general public or professionals (eg. clinicians, scientists). I will design the software system in such a way that it will be possible to extend the real-time health atlas to other diseases and geographical areas. Consequently beyond the lifetime of this project other researchers will be able to apply it to their diseases and geographical regions of interest potentially benefitting individuals, health care professionals and policy makers; creating long-term opportunities for improvements in quality of life and potential for reducing the cost of ill-health on economies. The 'professional' web application atlas could inform future research or health care interventions if unexpected temporal and/or spatial patterns are identified and aid the development of hypotheses to understand variability in disease risks; either evidence or lack thereof could potentially convey important information to the public health and policy authorities.

The real-time health atlas is exemplified by chronic kidney disease (CKD) in Salford. It will potentially be of benefit to those with CKD, or those at risk of developing it. Raising awareness of CKD could encourage individuals to make life style changes especially if supported directly by their health care professions and indirectly by policy makers (who could perhaps in the future implement screen programs to proactively identify early stage CKD). Money channelled into targeting people at risk of CKD could delay or prevent the onset of the later stages of CKD and hence other disease complications, improving people's quality of life and reducing the economic cost of this disease on the NHS and country.

ii) Research Activity B: MSc project

The immediate beneficiaries of this research will be the research and clinical communities (e.g. through journal publication) however once this research has identified the key risk-factors which determine the rate of kidney disease progression this knowledge should in the longer term be a step towards aiding better health care strategies and treatments benefitting patients, their families and perhaps reducing long-term economic costs to the country.

During my MSc project I will work in a clinical environment by applying advanced statistics to a specific kidney disease research question, this will benefit me by exposing me to a clinical setting and benefit the renal clinicians by exposing them to advanced statistics. Undertaking cross-disciplinary research to the point of producing research output (e.g. journal publications) will benefit both research communities by building strong collaborative links and an understanding of each other's skills, limitations and environments.

iii) Fellowship training and beyond

During this proposed fellowship I will undertake an MSc in statistics with a bias towards medical statistics and train alongside experts in health informatics and statistics. This will put me in a strong position at the end of the fellowship to take up an academic position at the interface of health informatics and biostatistics, where there is currently a skills shortage in the UK. I will then be in a position to lead my own research team at this interface and support the training of the next generation of researchers.