2111 E Highland Avenue, Suite 145
Phoenix, AZ 85016
Copyright 2021 Accumen
Never miss the latest in Healthcare Performance Insider by subscribing HERE!
@Copyright 2019 - Healthcare Performance Insider a publication from Accumen, Inc
While there has been much speculation about AI being the future of clinical diagnosis, the reality is we are still many years away from that outcome. Our conversation with Dr. Jonathan Chung of UChicago Medicine revealed the more immediate and long-term ways AI could impact radiology in 2020.
“When we think about the workflow in radiology in particular, AI has a near immediate ability to impact in three critical ways: improving efficiency, reducing waste operationally and most importantly, positively impacting patient safety.”
Leveraging computer aided diagnosis as the sole means to diagnose diseases is still many years from technical capacity and decades away from being culturally accepted. What would be of greater adoption immediately would be using AI to help rule-out significant abnormalities; that is, essentially using AI to identify studies which are highly likely to be normal or non-actionable. Indeed, in other countries, there is talk of using AI to identify these low-risk studies in the annual screening setting such that further analysis by a radiologist would not be pursued. Moreover, using AI to compare previous scans from the same patient correlated to the patients clinical history, physical, and lab tests would improve accuracy of AI algorithms in detecting significant findings on imaging which could alter patient management.
Reducing Waste Operationally
Leveraging AI on the operations end of the radiology practice could streamline patient scheduling, automate much of imaging study protocolling, optimize room turnover, and maximize billing efficiency and yield.
Positively Impact Patient Safety
Historically in radiology practices the primary KPI is turnaround time. This KPI has a very defined ceiling and the pressure to meet the KPI has been shown to impact clinical diagnosis negatively. With AI playing an initial role in ruling-out abnormalities we’re creating a more meaningful KPI that Dr. Chung refers to as Actionable Turn Around Time. AI can identify imaging studies with critical findings, which can then alert radiologists that these studies require immediate attention, thereby improving efficiency of patient triage and management for those who are most at risk for adverse medical events.
While the need for radiologists cannot be overstated, providing AI as an aid for clinicians could indeed support a more efficient practice with less waste. In addition, with Actionable Turn Around Time as your KPI, there is a positive impact on the safety of patients.
The current operational model of reviewing scans in a “first in first out” (a methodology with its route in accounting which dates back to the 1700’s) manner leaves critical patients at a severe disadvantage simply because of the order the scans were received. Leveraging AI as a tool to optimize the PACS worklist (placing studies with higher likelihood of significant disease at the top of the list) is an obvious use case, which is currently being explored by AI developers.