Clinical Recommender UI: Paper/Figure Revisions Figures - Due: 11/22/19

[JonathanChiang]

• [ ] include data on either resident trainees compared to more experienced physicians or even more specifically break it down to intern, resident and above. As the author mentioned, this tool may have specific impact on newer trainees or even older physicians more entrenched in their diagnostic pathways. Such analysis would be helpful to the community.

• [ ] If new evidence changes the standard of care, a healthcare system may manifest those new guidelines as new order sets or alterations to existing order sets. A recommender system based on historical data may continue make recommendations that areno longer indicated or even contraindicated. In fact, the system may reinforce old habits, so to speak. Similarly, new drugs will be less likely to be recommended.

This “cold-start” problem limits the utility of the approach. I’d like to echo what was said by surveyed providers, that there are certain scenarios for which this tool would work especially well. As the authors mention, there are significant barriers to implementation. Real-time extraction of granular clinical data and the ability of a custom CPOE system to interact seamlessly with the EHR-native CPOE system appear challenging.

review from interviews suggest that having a champion increases efficiency and up to date feedback on orders

[JonathanChiang]

• [X] Altered Scripts to account for resident and non-resident metrics.

[JonathanChiang]

See working directory with paper files at link below.

• [ ] Create some alternative tables / figures that address the reviewer comment about results for different user training levels (resident vs. attending).

• [ ] Start drafting responses to the reviewer comments in the document below.

• [ ] Aim to prepare them enough for resubmission consideration by end of next week.

[JonathanChiang]

• [ ] Table 3a: Table of Cases
• [ ] Table 3b: Table of Cases
• [ ] Survey Split

[JonathanChiang]

• [ ] Develop Script to pivot data to mirror Jon's work

[JonathanChiang]

• [X] Split the data between resident and non resident and saved to box