4 Lessons Healthcare AI Can Learn From Electronic Health Records
Originally published in Forbes.
Tremendous excitement. Grandiose predictions. Massive expectations. While these words describe today’s healthcare AI zeitgeist, they also describe how many felt about electronic health records in the 2000s and 2010s.
Since then, almost all U.S. health systems and medical practices have implemented EHRs, improving healthcare in some ways and worsening it in others. The outcomes have varied; organizations that invested in their workforce and systems have done better overall.
AI is the next phase of healthcare’s decades-long digital transformation. Although the rollout will differ, organizations approaching AI would be best served by applying lessons learned from implementing and utilizing EHRs.
Lesson 1: Set Realistic Expectations
Following decades of hope and hype around digitizing healthcare, many expected that EHRs would make healthcare safer, less costly and more effective. Things have turned out differently.
EHRs have been a mixed bag. For one, they empower clinicians with information yet overwhelm them with junk and nonsense. For example, clinical notes are now legible and easy to retrieve but often bloated with unnecessary, duplicative, and, at times, unintelligible content.
Similarly, EHRs bring clinicians and patients closer together while pushing them further apart. While portals make it easy for them to communicate between visits, obtrusive screens and keyboards in the exam room interfere with human connection.
EHRs make clinicians more productive in some ways yet less productive in others. For example, while it is easy to prescribe medications and communicate test results electronically, clinicians must process countless alerts and notifications.
Lesson 2: Put People First
Many have criticized EHRs for serving billing needs more than enhancing clinical care. As such, nurses and clinicians often find EHRs hard to use, contributing to burnout. Yet organizations that prioritized their workforce — for example, by communicating clearly, investing in implementation and personalizing training —— have done better.
With AI, organizations must start by winning back the hearts and minds of patients and healthcare workers who no longer believe in the promise that more technology will necessarily improve healthcare. This will require using AI to improve outcomes and experiences (not just billing and efficiency), making AI tools easy to use, and supporting change.
Lesson 3: Improve Systems Of Care
Health IT does not work in isolation. It becomes part of a socio-technical system involving various teams and workflows.
When implementing EHRs, many organizations digitized existing paper-based workflows and kept their teams unchanged instead of updating them for a digital world. This resulted in many misfit and wasteful workflows, often forcing healthcare workers to find workarounds and perform tasks that others previously did for them. However, organizations that have redesigned workflows and reconfigured teams for a digital world have done better.
Organizations must avoid making the same mistakes with AI. Bill Gates explained, “The first rule of any technology used in a business is that automation applied to an efficient operation will magnify the efficiency. The second is that automation applied to an inefficient operation will magnify the inefficiency.”
So, instead of rushing to automate broken processes or using AI as a band-aid for poorly designed technology, organizations must first optimize their EHRs, streamline the work and eliminate wasteful tasks. Initiatives like the Getting Rid of Stupid Stuff (GROSS) program can help.
Lesson 4: Continue To Invest In The Change
Many organizations treated EHR implementation as a one-time event, failing to recognize that it was impossible to fully anticipate what the “live” EHR would look like before putting it into production and training their workforce all at once.
Consequently, many EHR tasks are onerous (e.g., physicians at one health system must click 61 times to place a Tylenol order), and many clinicians do not use powerful EHR features (e.g., Epic reports that only one in three physicians use its chart search). Conversely, organizations that prioritized ongoing training and EHR enhancements have done significantly better.
The point is that with AI, the implementation should never end. Organizations should continually monitor AI, evaluate its effects, support front-line staff, and ensure that AI-based tasks remain in step with the work to be done.
The Stakes Are Too High To Fail
Healthcare organizations could harness AI to make care more accessible, effective and efficient. Yet success is not guaranteed. Those who draw on lessons learned from implementing EHRs — setting realistic expectations, putting people first, improving systems of care and continuing to invest in the change — are the most likely to succeed.