This article was written by Sally Embrey and originally appeared on the DataRobot Blog here: https://www.datarobot.com/blog/integrated-healthcare-means-integrated-data/
“Integrated healthcare” has become a bit of a buzzword as of late, but no one has settled on just one definition for the term. A 2016 paper that sought to explain integrated care, laid out four different definitions and five different conceptual frameworks. But if you break down these various definitions—and the hundreds of others that are out there—integrated care is collaborative and connected care. Or, put another way, integrated care is seamless patient care to the point where patients don’t feel friction when working with different provider or payer entities, no matter how many are engaged in their care.
Most people who have utilized the U.S. healthcare system know that friction tends to be part of the process, and they like the idea of an integrated care system. So, how do we make integrated care a reality?
The first step is integrated data. If data flows seamlessly between different entities managing a patient’s care, providers are in a better position to provide the best care possible. Within several years of Electronic Health Records (EHRs) being pushed by the Obama administration in 2009, 78% of physicians reported that EHRs enhanced patient care and 65% reported the records help identify potential medication errors. And while EHRs are just the first step, their implementation underlined an important issue—that providers can’t each work from a different dataset without making different decisions that could ultimately impact outcomes. Integrated data allows for a complete picture of patient services and needs.
And once the data is in one place, a new challenge emerges:harnessing its power. Big, complicated datasets require the right tools, and machine learning is critical to uncovering the trends hiding in these new datasets. Luckily, health professionals are increasingly aware of the importance of machine learning and artificial intelligence within their business. A December 2021 survey by Optum reported that 85% of healthcare leaders say they have an AI strategy and 48% have implemented it, an increase since reporting in 2020. These numbers will continue to increase as healthcare datasets expand and machine learning is further adopted into both operational and clinical decision-making.
While integrated care models have already been successfully deployed across the U.S., usually in the form of voluntary Accountable Care Organizations, we can also look to the UK as a potential role model with their development of Integrated Care Systems (ICS). ICS are soon-to-be mandated partnerships that will bring together various stakeholders and providers of the National Health Services to combine care across different hospitals and community-based services, including physical health, mental health, and social care. Starting in 2022, the ICS aims to improve population health metrics and decrease health inequities.
The British Medical Association states that better use of data and innovation is important for the ICS to be successful and that data sharing is a core focus. The data that comes out of the ICS has the potential to allow for both population and patient-level analyses that have previously been impossible due to lack of data sharing and data disparities. But visibility alone is not innovative enough. These types of analyses will require powerful data tools that are accessible and explainable to clinicians and researchers. From the other side of the pond, we should watch with keen interest as the ICS unfolds since we will be able to learn from their challenges and wins, especially as the call for integrated care in the U.S. grows louder.
It’s no wonder that people are excited about the idea of integrated healthcare with its promise of a seamless patient experience, lower costs, and improved outcomes. DataRobot is excited, too. While integrated care remains a buzzword, we can already show what AI can do when it’s implemented by medical and research facilities. Check out some of our current use cases. We will continue to unlock the potential of widespread data integration and adopted and deployed technological resources like machine learning.