Skip to main content


Recapping Our “Modernizing Patient Engagement” Roundtable Discussion

Patient Engagement 2

Perficient and IBM recently partnered with BWG Connect to conduct a survey of 100+ healthcare professionals exploring technology adoption in the post-pandemic environment. Experts from these organizations joined a roundtable discussion to review what is top of mind for healthcare organizations in 2022 based on the responses to this study.

The panel was moderated by Aaron Conant, co-founder and managing director of BWG Connect, and a panel of experts that included:

A Focus on Equity in Healthcare

Fowkes shared that in a study IBM conducted with 39 Blue Cross Blue Shields, equitable distribution of health was top of mind in every single organization. Similarly, participants in the BWG survey frequently brought health equity into their answers. Our panelists took some time addressing concerns around the use of automation furthering this issue:

Brendan Fowkes: The respondents very quickly jumped into a discussion of health equity, equitable distribution of health care, and where the social determinants fit in. What did we see over the past 18 months? Our most vulnerable took the worst beating.

Tom Lennon: As we start to push into AI and leverage AI, just because you’re using information around social determinants, it doesn’t mean that you’re going to drive equity. This is because the bias is inherent within AI until you can manage it and normalize. That bias is still going to be there. Learning how to use that and adapt it is important because that gives us the opportunity to help address some inequality. There’s also the opportunity that it could make it worse because AI doesn’t have the same feeling and natural intelligence as we would have when we’re worried about different groups within our population.

Eric Walk: Yeah, that’s where it’s really important to think about the tools you’re picking. Some AI tools have capabilities that help you identify bias in your models and others don’t. Those priorities are critical as you’re looking at the tools you’re using and the way you’re designing your models and thinking about building models. Another critical priority is educating your data scientists as you’re training your people. People build the models. People build the algorithms. People train the algorithms. It’s critical to train those people to think about these things and to be concerned about them and look out for the warning signs that there’s an issue in a model that’s exacerbating historical inequality.

EXPLORE NOW: Diversity, Equity & Inclusion (DE&I) in Healthcare

Other Key Takeaways from the Research Study

The survey generated interesting data points around data privacy, communicating ROI, and gradual technology adoption, which the panelists also explored:

Privacy, Compliance, and Trustworthy AI 

The research study revealed that many healthcare professionals had questions and concerns about the safeguarding of protected health information (PHI). The moderator opened this discussion with a question to the other panelists about protecting sensitive data when using automation.

BF: We use the term “trustworthy AI” because, if there’s any violation of a patient’s trust, you’ve lost everybody, whether it’s the nurses, the call center agents, or the patients themselves. We start with that foundation and a core belief that everything has to be trustworthy from the start. There’s no hidden “whatever else.” You’re not borrowing the data to go do something else.

If you start with that foundation at your core belief, you can start to overcome any of these objections. You’re going to design a system with the proper security, privacy, and consent. Borrowing a little bit of what we’ve done on the academic side, you can use other examples to make sure that everything’s secure when you’re designing that trustworthy story.

EW: And how are you going to control the data? There are CRM options out there that have the proper certification security controls in place to allow you to load your PHI and secure it and control it appropriately. Even so, that may not be an option for any given organization. How can we take the AI and the machine learning models, and bring them to where the data is, so we have fewer concerns about shipping the data to other places for training and for executing those models? We can more tightly control how that data is being sent, where it’s being sent, and how it’s being used.

You’re going to have more challenges in terms of just making sure you understand the way each of those solutions controls and secures PHI because a good number of them are compliant in various ways. Or you can look to other options that allow you to maintain trust through the whole chain of custody of the data by using more complex and nuanced solutions.  That allows you to take your modeling and do it where and when you need it.

TL: Yeah, that governance is important, especially as you put it where we’re sending that data. You must make sure that we’re able to limit the exposure in one place and send the information so it doesn’t have exposure all over the place. Trying to manage that when it gets throughout your entire organization can be quite tricky.

READ NOW: HIPAA Compliance and Protecting PHI

Communicating ROI When Consolidating to One CRM 

The panelists anticipate that the 81% of survey participants who see the value in consolidating to one CRM may have trouble calculating quantifiable value in doing so. They then discussed several ways that this 81% can effectively communicate ROI.

EW: There’s an opportunity to save some money and optimize your operations. But it is a question of how realistic it is and how much you’re going to spend to get there. So, I think that it can feel like the ROI is too far away. But there are some interesting opportunities that come from replacing the black box technology that comes with some of the less fully functional and less fully capable, but industry-specific platforms, and looking at the more enterprise full-spectrum platforms. So, it’s going to be an interesting journey as folks go down that way.

BF: And there’s definitely value there. But it’s hard to quantify, as Eric was just saying. So, to the 81% that see value in it, our advice is understand articulating your savings by retiring legacy applications, like some of these more black boxed pre-configured things that lack flexibility, as Eric was describing. What are you paying a year in applications like those? That’ll be an easy way to get to an ROI.

What does success look like when you’re having these conversations? Having a definable metric, and if we’re doing call centers, there’s metrics there that are pretty easily and widely accepted: cost per call, number of calls, deflection, and average handle time. You can put some hard ROI around that. But other ones are sometimes a little harder.

Our recommendation and our experience is let’s agree what success looks like before you define the use case you want to chase. Our recommendation is to find what the outcome is you want to measure, because you can’t fix what you can’t measure. So, we need to measure to prove it worked and then that success will build on itself.

LEARN MORE: Driving Increased ROI with CRM 

A Crawl-Walk-Run Approach to Technology Adoption 

Many see adopting this technology as a daunting task and a huge undertaking. The round table moderator addressed this by asking the other panelists how they are breaking down the process for clients who don’t think they have the bandwidth to begin any work with machine learning.

EW: You can do this use case-by-use case. You can tackle one challenge in one area with this kind of technology and slowly expand it. The upfront cost of dealing with security concerns and ethical concerns is lower than you might expect. If your use case is sufficiently simple, dealing with the additional concerns of more complicated use cases becomes incremental cost. The overhead there is not going to be huge and it’s not going to sink your plan to expand out into other areas.

BF: I call it “crawl-walk-run;” other people use “land-and-expand” or “start small.” Whatever metaphor you want to use, there’s a way to take something simple. An example was middle-aged men not taking blood pressure medication. If you looked at socioeconomic factors in that model, and we predicted not adherence, it would ship them a three-month supply and give them a call to tell them it’s coming. It’s not a complex prediction. It didn’t take a lot of work. What’s the generic cost, about 30 dollars to keep somebody adhering?

There still is true machine learning in the model. We did a whole scatter plot where we can look at all these different techniques and pick the right model that was most accurate. But it wasn’t a complex use case. It was just a different way of using the data they had and then helping them automate something to drive a benefit at the same time. So, it is all about crawl-walk-run.

EXPLORE NOW: The Healthcare Executive’s Guide to Intelligent Automation

See More!

This research study revealed key insights on the adoption of automation technology in the medical industry. Our panelists discussed how your organization can take advantage of these technological improvements by gradually adopting, communicating clear goals, and remaining compliant. Learn more about what industry professionals are prioritizing in 2022 by exploring the published research study and watching the full on-demand recording.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Cassidy Rimmey

Cassidy Rimmey is a Marketing Coordinator in the healthcare industry at Perficient.

More from this Author

Follow Us