A CIO Balancing Act
Project management expertise is one skill CIOs can bring to companies. Agile project management isn’t a one-size-fits-all solution, but certain core concepts apply broadly. One such concept is “vertical” slices of work. Working in vertical slices delivers a constant value stream. One challenge for the modern CIO is balancing this agile approach, favoring small, incremental changes, with strategically assembling the building blocks necessary to be ready to take a leap when new technology presents the opportunity.
Mental Health Center of Denver, a private, not-for-profit behavioral health provider serving people of Denver, Colorado by focusing on strengths and well-being, is starting to see the payoff of this strategy. As we accumulate more and more building blocks, new solutions can be developed rapidly. Some of the new solutions promise to be game-changing, rather than just incremental steps to improved workflow.
In developing a technology stack, a CIO’s job is to consider not just how a particular technology might fit a particular project, but also how it integrates with other technologies and what future opportunities it might enable
Before I moved into health information technology, I had a career as a clinical psychologist. Technology was just a hobby for me at the time, but I saw its promise to truly improve clinical care. Those promises didn’t include dreams of SharePoint in the cloud, but sometimes the steps needed to implement big dreams have humble foundations. I wanted to help clinicians make better decisions about the care they provide -- how should they choose right treatment to deliver to each person at the right time?
In developing a technology stack, a CIO’s job is to consider not just how a particular technology might fit a particular project, but also how it integrates with other technologies and what future opportunities it might enable. Several years ago, the Mental Health Center of Denver needed to overhaul its intranet. We settled on Microsoft SharePoint, not because it was the perfect tool for our existing intranet requirements, but because of how it fit with our existing data warehouse and our planned migration to Office 365. As one team in the Mental Health Center of Denver IS department implemented the new intranet, our systems architect, Steve LeClair, led a research spike to determine what possibilities Microsoft Azure might offer.
LeClair’s research meant that when one of our clinical program managers asked for an iPad so her staff could more rapidly connect to their virtual desktops, open up our electronic health record and look up information when working in our community, we had a better answer for her. With Azure we could now push information from our EHR to clinicians’ phones by building a mobile “lookup” application. No longer would they have to go through all those steps just to look up clinical information -- we have an app for that!
Mastering how to move information from our EHR, Netsmart Technologies’ myAvatar, up into the cloud opened up another opportunity. Instead of only accessing structured information, such as medications, diagnoses, and appointment records, we could begin to tap into the wealth of knowledge stored in the unstructured clinical narrative data that is especially prominent in behavioral health records.
We now have 15 years of clinical notes -- spanning both our legacy EHR and myAvatar -- indexed in Azure. A simple front end plugs into a “Search Console” in myAvatar so clinicians can now quickly locate the informational needle in the haystack of clinical documentation. The search console still requires the clinician to know what they are looking for. We are currently working on using natural language processing and machine learning (ML) techniques to mine these rich data sources to achieve true clinical decision support.
Our first ML project involved increasing the accuracy of our access clinicians’ decisions on how to triage intake appointments -- are people suited for outpatient therapy (so therefore schedule their intake with a therapist) or would they better served with intensive case management services (meaning the access clinician should conduct the intake)? Mental Health Center of Denver Director of Information Services CJ McKinney used ML to classify people seeking services into these two categories. Dr. McKinney found that we could rely on just a single assessment question from our 27-item needs assessment survey and still improve classification from 76% accurate to 85%.
Future ML projects hope to address questions such as what elements of treatment are most important in moving people towards their goals faster, how can we predict risk of suicide, and ultimately, how can we anticipate what information a clinician should need to know, and present it to her proactively. Even if we never get all the way there, we still have what we already built. By taking the approach of vertical slices, our clinicians are enjoying many of these solutions right now.