By unleashing the power of machine learning, we can better understand behaviour, empower patients to make smarter decisions – and save billions of euros.
Unhealthy lifestyles are driving an explosion in chronic conditions, including obesity, diabetes and cardiovascular disease. By choosing to smoke, having an inconsistency in maintaining a healthy diet and opting out of exercising, we place ourselves at risk of ill-health. At the same time, some patients are neglecting to take their medicines as prescribed or are misusing antibiotics – with devastating consequences.
Around twenty–one percent (21%) of US healthcare costs are attributable to the influence of human behaviour. For example, poor medication adherence alone costs the US more than $100 billion annually. Harvard and the World Economic Forum have estimated that non-communicable diseases result in economic losses for developing economies equivalent to four to five percent (4-5%) of their GDP per annum.
A patient-centric approach to behaviour change promises not only to improve clinical outcomes, but to address the rising demand for health services. Better education and awareness can help individuals to make smarter choices. There are a range of interventions available, but the challenge is providing the right patient with the right behaviour change intervention at the right time.
If We Can Predict, We Can Prevent
Now we have new tools at our disposal, informed by research from psychology and behavioural economics, and powered by technological advances. As someone with a keen interest in behaviour change and the predictive power of analytics, I believe machine learning can help to make our health systems more sustainable. Artificial Intelligence (AI) allows us to evaluate how an individual makes lifestyle decisions and tailor behaviour change programmes to suit their needs.
When considering an example of poor medication adherence, if we are aware of who is at risk and can predict how they will behave, we can deploy timely and effective interventions that improve measurable outcomes and contain the redundancy in costs.
This involves a variety of analytical statistical techniques from predictive modelling, machine learning, and data mining, all of which analyse current and historical facts to formulate predictions concerning future events. These approaches provide a predictive score for each individual patient in order to determine, inform, or influence the processes for the adoption and change in a patient’s adherence.
Through machine learning platforms, hospitals have identified that costs were twenty-four percent (24%) higher for socially isolated individuals than for socially connected individuals with an equivalent level of clinical risk, and that the socially isolated individuals also had lower prescription drug use.
Such insights can help to identify high-risk patient subgroups before high-cost episodes occur. Interventions targeted toward these subgroups can then be designed accordingly, utilising remote and self-care technologies to support and empower individuals, and connect them to clinicians and other influencers.
New behaviour change programmes are based on a person-focused, rather than disease-focused, approach. Utilising this behaviour data will allow pharmaceutical and device manufactures to move toward personal and precision medicine.
At my company, we estimate that programmes designed under the new paradigm could deliver ten to fifteen percent (10 – 15%) reduction of costs in targeted populations, while offering productivity gains, better outcomes, and a better quality of life.
It is my belief, that the implementation of analytical person-focused paradigms offers measurable outcomes, and meaningful results for disease management programmes, and the much-needed reduction in medical costs.