An effective analytics provider promises to bring novel data sets and insights to the table, optimize workflows, build efficient processes, and deliver superior financial returns in evolving value-based payment models.
The Challenge for Buyers
The “analytics” is a complex, nebulous family of functionality and the marketing noise surrounding many applications these days can be extremely misleading. Predicting what the product even looks like post go-live and how clinicians will end up adopting the application is often hard to nail down. Analytics itself covers several applications that all hold immense promise for healthcare applications. For instance:
- Computer vision solutions can automate diagnoses,
- Natural language processing can transcribe documentation and obviate simple and non-value-added tasks, and
- Big data analytics can parse through an ever-growing wealth of data to provide personalized insights to providers on the frontlines.
Despite all its promise, analytics solutions can be extremely difficult to effectively implement, especially in a high-stakes hospital environment. While every provider may claim to be validated, easy to deploy, and employ equitable algorithms, the reality is different.
Software “Probabilistic” vs. Logical
AI-based solutions in healthcare represent a paradigm shift in how software is developed, deployed, and used in real-world settings. This kind of software is no longer principally deterministic. It is not programmed with precise, traceable, and relatively clear logic. Software that uses AI is “probabilistic” – i.e., documentation, reports, and care recommendations are all made through machine learning models and trained on unimaginably large data sets. It can be impossible to trace why a software recommended this product, or why facility A was benchmarked to facility B across the country.
As you start to understand what analytics is and how it can benefit your organization—that all changes. A high quality and comprehensive analytics solution can transform an organization. And capitalizing on those benefits starts with understanding what to look for when choosing an analytics provider.
Here is a checklist of things to consider.
Interfaces vs. Integrations: Why This Matters
When it comes to analytics providers, the waters between an interface solution and an integration solution can get muddied. An interface is a relatively simple connection between two existing systems (e.g., your AI vendor and your EHR) that inevitably causes some extra work or double-documentation for providers.
An integration, on the other hand, is a solution deeply rooted in the existing systems to seamlessly improve the workflow. The SMART on FHIR integration model is by far the most efficient way to embed third-party applications in an EHR. These applications are available natively inside the existing EHR and generally eliminate double-documentation and the typical headaches of loosely connected systems. Leading EHRs like Epic and Cerner have embraced the SMART on FHIR model by launching interoperability programs through the App Orchard (apporchard.epic.com) and the Cerner Open Developer Experience (code.cerner.com). Regardless of which EHR you use, the stronger the integration, the more potential the solution has to drive meaningful clinical benefit, without sacrificing the workflow.
And when a provider can’t deliver on the technical complexities of an integration solution, the marketing message may muddy the truly powerful AI that the product offers. When selecting an analytics provider, ensure it’s not just an interface, dig deep into the quality of the integration, do an extensive integration scoping exercise, and truly understand how things work together.
Today, clinicians spend almost half of their days typing or dictating documentation in the EHR .1 It’s been demonstrated that the ever increasing burden of documentation has contributed to burnout of physicians.2 With clinicians already spending a large percentage of their time with documentation, it’s imperative that an analytics solution subtract from the workload and not add to it. It’s important to ask yourself before selecting an analytics vendor – is the solution adding to documentation efficiency or is it actually increasing the time that clinicians have to work away from patients?
Documentation efficiency begins with the interface vs. integration discussion and moves into how the solution adapts and optimizes existing workflows. Some additional questions you’ll want to ask include:
- Is the AI component of the solution intended to eliminate tasks in the workflow?
- Do the AI models require clinicians to document new fields in the EHR, or navigate to a standalone system to see the output?
- Is the provider provided excessive data that looks helpful but could detract from their day-to-day worklist?
At the end of the day, key decision-makers have to make the call on which wound care analytics provider is best for their team. However, it’s not necessarily a call that has to be made completely on your own.
Put a heavy focus on the availability and quality of references behind each solution you’re considering. Are you the first client the company has had, or are there trusted references who can validate the quality of the solution? Is there anyone who works with your EMR that can speak to the ease, quality, and effectiveness of the solution?
And while there is nothing inherently wrong with working with an analytics provider that’s brand new to the space, be aware that there will be challenges along the way until the company achieves the critical mass needed to work through every possible EHR or integration configuration.
1Annals of Internal Medicine, Allocation of Physician Time in Ambulatory Practice: A Time and Motion Study in 4 Specialties, 2016