Alex Waddell, Chief Information Officer, Adobe Population Health Leadership.
Rarely are electronic medical records (EMRs) liked by those who use them. They are often seen as a barrier to care. This is due to two major problems: inaccessible data and the time it takes to chart accurately.
EMRs have a problem with inaccessible data due to the nature of data modeling and user interface design. Often, pertinent information is hidden deep in the chart, and clinicians—who may have five minutes before a visit to brush up on the patient’s history—are stuck fumbling around a system trying to uncover valuable information.
The time it takes to chart is also a difficult problem. No matter how advanced your EMR is, the clinician must still write out their note. They can fill out hundreds of assessment questions as drop downs and check boxes but must sit down and write a text summary of what was captured in those structured data points. They must also provide context in areas that cannot be captured in a drop-down, such as how they addressed an issue. This precious time that should be spent with the patient discussing how to solve complex medical and social challenges is often rushed due to a need to provide the organization with documentation. Those who do not limit time with patients are left charting after the visit to catch up, frustrating the clinician and opening the opportunity for charting mistakes.
So, how does artificial intelligence (AI) help solve this problem? The answer is by using natural language processing for summarization. Imagine a world where a clinician accesses a chart and there is already a summary of the patient’s medical history waiting for them, which could stretch across months or years of visits with that person. For those who have the time to deep dive into charts before the visit, this extra time can be spent preparing to see the patient. For those who do not have the time and therefore miss out on the historical context of the patient’s chart, they now have insight into critical data that may change the direction of the visit for the better.
Once a visit is complete, the clinician could trigger a natural language model to collect essential information and pre-fill much of the notes for them. This leaves the clinician with the easier task of simply editing the prompt that the AI created and adding any missing information.
In the healthcare industry, we have spent the last decade creating paper assessments, charts and EMRs to gather and maintain valuable data insights, but in the process, we have created unworkable systems for those on the front lines. As technologists, we must not forget those we build systems for and ensure they help clinicians provide quality care.
In doing this, organizations can reap significant returns in opportunity gains. That may be fitting in another visit or two due to time saved or enabling more informed decisions that could positively impact a patient’s life and ultimately reduce costs for the patient and healthcare system.
When considering AI use cases in your organization, do not lose sight of the low-hanging fruit. AI does not need to be complicated or expensive. Start by looking at how you can drive efficiencies within your organization, giving time back to those who do the work.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?