Utility of amazon-inspired algorithm for resident procedure logging
Bacharouch A, and Goyal N. Utility of amazon-inspired algorithm for resident procedure logging. Academic Emergency Medicine 2020; 27:S238.
Academic emergency medicine
Background and Objectives: Accurate procedure logs allow residents to demonstrate procedural competence and meet accreditation requirements. Residents often perform multiple procedures on the same patient but may only remember to log a single primary procedure, or they may forgo logging some procedures due to time constraints. To mitigate this, Henry Ford Hospital Emergency Medicine (HFHEM) developed two logging tools that recommend additional procedures to record when a primary procedure is submitted. The first tool (“Website”) provides suggested procedures based on a static linkage list predetermined by residency leadership. The second tool (“App”) uses an Amazon-inspired algorithm to provide dynamic suggestions based on selection patterns of other residents. For example, the App would say “Residents who logged 'I&D' frequently logged 'Local Anesthesia' or 'Ultrasound'.” Our objective was to determine whether the dynamic algorithm leads to a greater frequency of procedure co-logging compared to the static linkage list. Secondarily, to determine whether such suggestions successfully prompt residents to log procedures which they may have otherwise forgotten when using traditional logging tools.
Methods: Procedure logging data at HFHEM for academic year 2018-2019 were retrospectively analyzed. The rates at which residents co-logged 1, 2, or ≥3 procedures using the Website or the App were compared using chi-square statistics.
Results: 8,656 entries were logged: Website 6,804 (78.6%) and App 1,852 (21.4%). The rate of co-logging >1 procedure was significantly higher with the App as compared to the Website (48.3% vs. 31.1%; p < 0.001). Similar results were seen for co-logging ≥3 procedures (16.6% vs. 12.9%; p < 0.001). Overall, 34.8% of submissions had at least 2 procedures co-logged.
Conclusion: The Amazon-inspired algorithm was superior to the pre-determined static list in promoting procedure co-logging. Moreover, suggesting procedures (regardless of the algorithm used) led to a high rate of co-logging. While some technical expertise is required to implement these tools, the source code is available to all at no cost. This innovative algorithm may decrease the time needed to log procedures and may improve the accuracy of the record by capturing procedures that may have been forgotten when using traditional logging tools.