Hurdling four barriers on the road to predictive analytics

Data shares many of the same qualities as a magic 8-ball and a time machine. Understand it and you can predict the future; use it effectively and you can get there faster.

Healthcare organizations now possess massive quantities of valuable data, and many are wondering how to extract actual return on their investment. Unfortunately, it’s not exactly easy. Indeed, at the recent Becker’s Hospital Review 3rd Annual Health IT + Revenue Cycle Conference, Andy Bartley, senior solutions architect at Intel Corp. identified four common roadblocks healthcare organizations face in adopting predictive analytics.

I recently joined BaseHealth – a company that takes a different approach to population health. We make patient and population data immediately actionable through the use of predictive analytics to enable physicians to deliver better care and reduce healthcare costs. We’ve met each of these four roadblocks during our journey, and we’ve been able to cross them and achieve great results in partnership with our clients.

1. Scalable infrastructure
Scaling analytics is a giant hurdle to cross. Table stakes are to have data integrity, be cloud-based and ensure HIPAA-compliance. Beyond that, one key is to optimize data access and visualization. Done effectively, it will be easy for users to locate and understand complex information on their own.

A second key is to connect to all sorts of data sources. As data-mavens, we're responsible for scooping data and making it understandable. This can be done through creating application programming interfaces (APIs) or purchasing them through a vendor like MI7. True change and actionable intervention plans can’t come from siloed data sources, so we combine real-time patient data with our causal expert model that is based on the peer-reviewed medical literature and assesses risk for 43 diseases and conditions. BaseHealth also uses unstructured, “dirty” data in our models. With data flowing in at this volume and pace, there’s no time to rebuild it to fit a specific criterion.

2. Executive sponsorship
Depending on their role, executives within a healthcare organization focus on different objectives, and have various metrics to measure success.

To receive executive sponsorship and cross this barrier, predictive analytics teams need to speak the language of more than one executive. It’s important to address clinical outcomes, of course – but tying those results back to financial outcomes and showing a direct, positive return on investment is key. If the C-suite sees predictive analytics not as an expense but as a cost-saving opportunity, the technology will be on the fast track to adoption.

3. Use case selection
Mr. Bartley suggests a great strategy of “balancing quick wins and big bets.” By showing positive results on short timeframe projects, predictive analytics teams can do more ambitious work on longer-term projects in the background.

At BaseHealth, our technology is able to provide a much clearer picture of the risk that healthcare organizations are taking on when becoming accountable for a given population. This forecasting and visibility is a quick, tangible benefit for clients during contracting efforts. Our longer-term success, though, is in uncovering and giving health systems the information they need to intervene with a group we call the “Invisible Patients,” those who appear to be healthy on paper but will soon enter a downward spiral of negative health issues. Helping care managers reach these patients and prevent costly conditions will have a much larger impact on an organization’s bottom line.

4. Change management
Here’s an inconvenient truth: people generally want to keep doing the same things they do every day – it’s their workflow. That makes change management difficult! I’ve found that instead of attempting to change the workflow of a healthcare organization, a better approach may be to optimize that workflow.

At many healthcare organizations, care managers have a list of patients – usually the sickest – and they call these patients regularly. Unfortunately, they usually don’t have the benefit of understanding those patients in greater depth than what their EHR data allows. Using predictive analytics, care managers get a different list of patients: not those that are the sickest, but those that will become the sickest in the near future. Care managers want to help patients get better and there isn’t a group of patients more in need of their help than those who are on the verge of facing a major adverse medical issue.

Instead of creating a new workflow, we optimized the one that already exists. And when this is successful, it can catalyze even greater change management and willingness to adopt a new approach.

Copyright © 2024 Becker's Healthcare. All Rights Reserved. Privacy Policy. Cookie Policy. Linking and Reprinting Policy.

 

Featured Whitepapers

Featured Webinars