The Australians may have big ambitions, but we can match them, as long as we don’t over play the role of big data. By Dr Marc Farr
Cricket and the Ashes aside, there are other reasons to be keeping an eye on Australia. The Australian Digital Health Agency has been making significant strides with MyHealth Record, most recently announcing a partnership with the Consumers Health Forum of Australia to ensure Australians have a real voice in developing its digital health capability.
This comes on the back of Primary Health Care, Australia Clinical Labs and seven other software vendors and labs signing agreements with the agency.
The agency would appear to be on track to deliver its target of delivering a My Health Record for all Australians in 2018. So, can we match this initiative in terms of ambition? If you take into account the UK government’s aim for greater use of artificial intelligence in healthcare delivery then the answer could be yes.
The government wants the UK to become the best place in the world for developing and deploying AI to start, grow and thrive. A review by professor Dame Wendy Hall and Jérôme Pesenti highlighted the estimate that AI could add an additional £630bn to the UK economy by 2035.
If this gives you a familiar feeling that we could be on the cusp of some significant changes in the way healthcare is organised and delivered, you are not alone. There is a growing awareness of the benefits that AI can bring for healthcare providers.
From healthbots that help with emergency care triage to the analysis of large amounts of health data. Some of the drivers for these changes in healthcare are: increased volumes of data, the growth in the number of experts with the specific high level skills and the availability of increasingly powerful computing capacity.
The potential of big data analytics and predictive analytics is often talked up at this point. The suggestion is that modern day computing power, combined with the falling price of data storage, means big data analytics is becoming more achievable.
Predictive data analytics to reveal current patterns in disease and risk leads to general excitement about the potential to save money and improve health outcomes. In the promised land there will be sophisticated, animated simulations of hospital activity and wider population health management.
However, big data should not be getting all our attention. Our focus now should be on small elements of relevant data – finding the information that will help us target resources means we have to be selective. Our energy is best spent on collection, cleaning, enrichment and analysis of data to determine its relevance.
For now, we should be analysing a multiple of small, simple but accurate surrogate measures of performance. Take the four hour emergency target for instance. While it survives as a key target, we are not seeing widespread use of standardised models that predict flow well. “Stranded” (length of stay of over 14 days) and “super-stranded” patients (where length of stay is over 21 days) are being predicted well in some trusts, but not in others.
We should be collecting these metrics and aligning them with indicators such as: the number of ambulances that have arrived in the last hour, the number of delayed transfers of care compared to a rolling average and the number of available beds currently in clinical decision units, the number of mental health patients requiring psychiatric review.
Getting accurate and available data in real time and concentrating on maximising the use of them is what we should be doing to prepare the ground for big data projects.
As for data sharing, we don’t necessarily need large scale data sharing initiatives, just safe places for people to work with health data at a local level. We should be able to allow technology/AI companies to test their apps and solutions in a safe and secure environment. We can make sure all the appropriate safeguards are in place and give them the chance to make mistakes before anything is rolled out on a national level.
Hopefully, the Sustainability and Transformation Partnerships programmes around the country can provide the political relationships for this type of collaboration locally. In Kent and Medway, for example, there is exciting research to link pharmacy, primary care and acute data in order to improve drug compliance after stroke through better collaboration and signposting with patients.
This will give us tried and tested solutions which can help to realise some of the benefits of technology and AI that we hear so much about. If data is the new oil, then we need the pipes, the oil rigs and the engineers. They’re not here today but they’re coming.