The Role of Big-Data in Addressing the Determinants of Non-communicable Diseases in the Ageing Era
Sydney University hosted the World Universities Network Research Open Day at the Nanoscience Hub yesterday.
The theme this year was big data. The specific context was how it can be used to address challenges in the health.
I am still processing much of what was presented. Here are a few:
- Before sharing visualisations showing how healthcare ‘snow-birds’ migrate south from New York to Florida for the winter, Professor Martin Zand (University of Rochester Medical Centre) showed us US state borders depicted by frequency of visitation by users in a state administered medical care program. He also showed us how to track the original colonial terrorists, using data.
- Professor Andrew McLachlan (University of Sydney) talked about disruption in health and the need to brace for that level of change in order to deliver the improved health outcomes that can transform the industry.
- Dr Michelle Morris (University of Leeds) shared a startling fact: DYK 70% of deaths, globally, are from non-communicable diseases? Of those, Dr Morris described the evolution of big data practices, so that researchers now consider big data as a challenge to find new forms of existing data, rather than the Vs (volume, velocity, variety, veracity). Efforts to link supermarket loyalty card data to health records open up potential to unlock deep insights into the diet of individuals, on a large scale. GDPR means that individual consent will be needed to link specific data records. Balancing competing priorities is a sensitive issue.
- Professor Alistair McEwan (University of Sydney) introduced me to the Cerebal Palsy Alliance (an organisation created by families of those who are diagnosed) and showed how understanding the electrical properties of the body could make a difference for those with this condition.
- Professor Chin Hur (Columbia University), gave one of the most big-brother presentations. Using his machine learning technique, he was able to pinpoint which nurse undertook which task in the patient care cycle. No personalised data required. His research looking at clinical predictors using an endoscopic database (not a happy place to find yourself if you had an experience that was listed in this data set at the outset — but made better or worse depending on the care you received) measured, with incredible precision a range of factors that could transform patient care. Real time insights, monitoring and diagnosis may not be far from patient care methods.
- Professor Clara Chow, Academic Director Westmead Applied Research Centre, had one for the technology enthusiasts. Her team delivered a text messaging program that delivered structured text messages relating to general health, diet, exercise and treatment to a group of patients and measured the outcome of that support service against a control group. The results were significant. But — as a clinician trying to deliver care to her patient group, her PhD student did an audit of the 278 apps in the app-store that claim to deliver results of the kind she was aiming for. Only 1 of that group passed basic compliance testing against the standards of care that applied. Could that be an invitation for some kind of health care stamp of approval for apps being listed in the health section of the app store?
- Getting deeper into neural networks, Associate Professor Kelvin Tsoi showed us how machine learning could indicate hypertension, on a patient level. I’m still processing this discussion.
- Professor Michael Valenzuela had a voice recognition application that tested for dementia in patients — without human intervention. Like Dr Chow, he is using technology to administer basic tests and treatments, and also cited compelling statistics about the number of apps in the app store that make unmerited claims about efficacy. His application took account of a range of factors that need to be considered in arriving at a reliable measurement of memory, and outlined a robust method used to built that model. Clinicians can use the app to test patients, as frequently as they like. Patients can self test. In an interesting angle on calls for collaboration, Professor Valenzuela mentioned that the app has been developed for english speaking patients only — he is interested in finding data sets to adapt it for patients across Asia.
- Associate Professor Jinman Kim, Research Director at Nepean Telehealth Technology Centre at Nepean Hospital shared some of the applications developed through collaboration between clinicians and engineers, to improve patient care. He made the point that health has become digital. Projects small and large, simple and complex are being shipped within his program — and he seems very enthusiastic to get more lined up.
- Combining a dose of history lesson and a fun fact about bees (is there ever any end to the interesting things to learn about bees??), Dr Nicholas Fuggle (University of Southampton) showed us the Hertforshire Project and all the interesting things that can be unlocked because of the fastidious births-record-keeping efforts by a team of British midwives in the 1920s.
- Assistant Professor Zhiyao Duan (The University of Rochestor) took us through neural networking and deep learning in the diagnostic process. The potential to use technology for diagnosis is off the chart. Development in this field is because electrical and computer engineers are meeting the health care workers and the patients where they are, and building technology that improves outcomes. We saw a demo of how his technology learned to decipher between background static and key messages, as just one example of what’s possible. A quick look at his bio and you can see his data sets are wide and varied.
- Dr Jill Newby (University of New South Wales) is a psychologist. Her digital, self-guided treatment for depression and anxiety in cancer patients has shown positive results, both as an accompaniment to in-person treatment and as a standalone service. It’s a dynamic application that can be customised for patient status and the results show an impressive reduction in anxiety amongst users. Dr Newby made a point that often comes up in technology discussions, in noting that some of her colleagues were wary about using the tool, expressing a concern that they perceived it as a substitute for their services. It’s a challenge that affects every profession and trade and we need broad scale solutions to address that concern.
- Ms Rachel Oldroyd (University of Leeds) has a message for dodgy curry houses. Big data can find you, regardless of whether customers who get food poisoning from your product report to the doctors office for help. Data analysts can use consumer generated data for public health surveillance. Foodborne illness was the subject of her study and some of the data sets available are big.
I’m still processing so much of what was shared yesterday.
What impressed me vividly was that this group of researchers, who are breaking ground in their fields, want to work with those in adjacent industries and fields. They are sharing discoveries and methods with public health administrators and others to improve wellbeing — whether that is for an individual patient or society at large.
Learning the language of those participants is part of the challenge in improving health outcomes. Almost every presentation invited potential collaborators to come forward, many of them looking for people with technical, data and digital skills.
The challenges are broad and the goals are ambitious. The World Universities Network has identified 4 themes which are the focus points for that group, being:
- response to climate change
- understanding cultures
- public health
- global higher education and research
Perhaps Professor Zand’s formula for effective collaboration is helpful for those considering reaching out. He prescribes a great deal of faith, doubt and determination to make it work.
So, time-is-a-wasting. Let’s get to work.
