ETIH: Data science allows for targeted prediction and intervention that saves lives

Return to eHealthNews.nz home page

Guest column from Emerging Tech in Health Speaker Kevin Ross

Imagine if an algorithm could save your life rather than convince you to purchase a product or vote for a political candidate. Imagine a world where we could predict the likelihood of having a stroke, or developing heart disease or diabetes, and put in place an intervention. We can, and we are.

Just last year a Wellsford man had an ultrasound after being alerted by a pre-screening trial that he was at risk of the silent killer, aortic abdominal aneurysm. True enough, on further investigation by his doctor, he was found to have a 6cm bulge in his abdominal aorta. A four-hour surgery saved his life.

This life-saving trial is part of a data analysis project carried out by Precision Driven Health – a research collaboration between Orion Health, Waitemata District Health Board and the University of Auckland – to develop algorithms that save lives by using risk criteria to precision screen potential patients.

Helping a system in crisis

With recent reports of district health board budget blowouts, medical staffing at a crisis point and winter looming, our hospitals are under immense pressure. Already this year clinicians are feeling the pressure of demand, with Middlemore Hospital’s emergency department experiencing the highest number of patients ever in January. Over the past five years, emergency department attendances across Auckland’s three DHBs have increased by nearly 19%. The emergency department is the ambulance at the bottom of the cliff, the least health-effective or cost-effective way to treat a patient, and sadly, sometimes it’s too late to treat. The costs to treat our communities are mounting, and the system is close to breaking point.

But some demands on the healthcare system are for good reasons. We are living longer and coping with chronic conditions as we age. However, a large portion of the demands on our healthcare system is avoidable. It is no secret that hospital admissions and specialist services could be better utilised in the community.

We see similar patterns in healthcare systems globally. Global statistics point to more than a third of all expenditure in the healthcare system each year being avoidable. It is likely to be about the same in New Zealand’s system. Most of this waste is attributable to giving care when it isn’t necessarily needed, or missing opportunities to put in place interventions that help circumvent the need for more intensive care.

A fundamental problem healthcare organisations face is the lack of analysis and insight available to help them to predict and plan for patients. This is where data science can help. Data scientists are beginning to use machine learning to develop meaningful tools that can provide clinicians and patients with valuable insights that allow them to manage long-term care much more effectively and, fundamentally, keep people out of hospital.

Using historical health records and new data from genomics and personal devices, we can train algorithms to calculate the likelihood of almost any health outcome. This ability to predict allows families to prepare, clinicians to target their treatment, hospital administrators to plan, and healthcare systems to operate more efficiently and ultimately improve the health of our community.

Make models freely available

To enable new insights and treatment models, we need both data and translation to be in the hands of patients, their carers – both professional and personal – and the administrators of resources. Today, our clinicians can access a narrow range of our medical records, and patients can see very little. Precision health manages people’s health based on their characteristics and requires us to expand the data sources being considered, and to make models freely available so that the data can be understood.

The flow of information between healthcare professionals will help to better diagnose patients. Devices can help monitor heartbeat and detect arrhythmia, apps are helping to diagnose anything from the common cold to skin cancer. But there also is an opportunity to use data from multiple sources to better manage complex health issues such as diabetes, aortic abdominal aneurysm or post-stroke care. Data analytics can inform ways to deliver better healthcare, by identifying patterns in outcomes, risks and effective intervention and treatments.

While prediction is key to better understanding the future resourcing of healthcare, intervention will ultimately have the biggest impact on the resourcing of the health system. The benefit to hospitals is the ability to plan more accurately. The benefit to patients is receiving better care.

Read more about the 2018 Emerging Tech in Health Symposium.

Kevin Ross is the general manager of research at Orion Health and head of Precision Driven Health, a public private partnership.

Return to eHealthNews.nz home page