Written by Tracey Cotterill, MD Population Health Intelligence, Civica
Civica’s Tracey Cotterill explores how machine learning can provide the data intelligence we need to deliver better healthcare for everyone.
Prevention, as we know, is better than cure. Intervening early and stopping health issues from escalating leads to improved outcomes all round — for individuals, communities and a range of healthcare providers.
With our UK healthcare system under huge pressure, there’s a pressing need to shift to this more proactive approach. But to make it a reality we need to know what’s behind health outcomes. Why, for example, do people in a particular area have fewer healthy life years than those in another part of the country?
Population health intelligence is about finding the answers to questions like this, so we can make the right interventions at the best time. As integrated care systems (ICSs) work to tackle some of the NHS’s most pressing problems — such as health and care inequalities and financial sustainability — it’s a discipline that can help make the connections between health outcomes and the factors that influence them.
Because ICSs will bring together NHS, local authority and third sector bodies, they’ll have access to all the data needed to gain a deeper understanding of population health — everything from NHS records to data on education, housing and crime. The challenge, of course, is how to extract the relevant insights that can point to new and better ways of doing things.
Crunching the data
Data on life expectancy is a good starting point. With ‘levelling up’ in the spotlight, it’s also a sound measure of health inequality. Unlike health data or patient reported outcome measures (PROMS), life expectancy data doesn’t depend on people having accessed healthcare. So it’s a more inclusive and accurate proxy for overall population health. The variations in the data are stark: men and women born in Glasgow City today will live around 10 years less than those born in Westminster or Kensington & Chelsea.
We need to understand what’s behind these stats: to help us, machine learning can make meaningful analysis of disparate datasets — able to rapidly work across huge volumes and multiple sources of data to identify patterns that can guide decision-making. Population health intelligence can help us analyse the causes of death at different ages in different demographics and the wide range of influences on them.
Factors such as education and housing affect health outcomes. As an example, data analysis may connect high levels of poor housing stock with respiratory illness. This could ultimately show that making improvements to living conditions today could prevent people developing chronic conditions that lead them to depend on multiple health and care services in the future.
In a similar way, information on dental health — such as the number of people in a single area having teeth removed at a young age — could also be a predictor of chronic conditions such as diabetes and heart disease in future life, and so direct healthcare interventions towards support for diet and lifestyle change.
By connecting health data with environmental information, such as air quality or the amount of available green space, population health intelligence techniques could show local authorities where to focus their investment where people’s physical and mental wellbeing will benefit the most.
Intelligence for all
Combined with powerful data analysis, the rise in health-related wearables also supports the shift toward more personalised and proactive healthcare. From simple step counters and heart rate monitors to sophisticated continuous glucose monitors, people are increasingly willing and motivated to track their own wellbeing. When connected to healthcare systems and analysed by machine learning algorithms, wearable devices and apps could support preventive healthcare by alerting professionals to potential issues, for example an individual showing pre-diabetic symptoms.
The UK’s move towards integrated care systems presents a huge opportunity to build a proactive approach to healthcare based on insights gleaned from many different data sources. Machine learning, as explored in our latest Perspectives* volume is vital for unlocking this potential, helping to build more innovative, impactful and cost-effective healthcare models for everyone in society.
Originally posted here