#BMIweekendReads - ...The Case for Teaming Intelligence

New weekend read out of AI Magazine. It is on the idea that AI is interdependent with the humans who use it. As AI has gotten more complex, the human-AI interactions have gotten more complex as well (e.g., pilot training time has increased with airplane sophistication). The main takeaway is that AI should be designed from the start to support the management of interdependencies with people.

tweet reply to comment

#BMIweekendReads - Use of a Novel, Electronic Health Record–Centered, Interprofessional ICU Rounding Simulation to Understand Latent Safety Issues

Bordley et. al.’s paper in Critical Care Medicine (Oct 18), showcases the use of simulation to understand the recognition of latent patient safety issues in the ICU. The study assessed interprofessional ICU rounding teams as they prepared for and discussed two simulated patient cases.

They found that teams averaged nearly 70% recognition of the total safety issues, which is over twice the rate of error recognition of individual providers who assess the cases independently (32%).

Figure 1. B. Shows the rates of recognizing the different latent errors present in the medical record of the two simulated cases. It is interesting to see the differences in recognition among the different types of care providers (reinforcing the value of interprofessional care teams) and to see which safely issues are nearly always missed.

Supplementary Figure 3. Shows the average number of EHR screens viewed by each provider type. The large number of screens needed to be viewed once again supports the need for better EHR design. We need EHR’s that highlight the right data, for the right task, to the right user, and at the right time (for example).

Tweet reply to comment

#BMIweekendReads - Machine Learning in Medicine

This review article in the New England Journal of Medicine provides a vision for where Machine Learning in Medicine is now (Figure 1), where it is going (Figure 2), and how we get there.

I am encouraged by the Clinician Workflow section which calls for “Machine learning that…can help expose relevant information in a patient’s chart for a clinician without multiple clicks.” Such workflow efficiencies are precisely what a Learning Electronic Medical Record (LEMR; pronounced lemur) System is designed to do.

Figure 1 . Current State

Figure 1. Current State

Figure 2.  Future state (using Machine Learning In Medicine)

Figure 2. Future state (using Machine Learning In Medicine)

Tweet reply to comment

#BMIweekendReads - A method for the analysis and visualization of clinical workflow in dynamic environments

A new paper out of Vimla Patel’s lab provides a nice summary of methods for analyzing and visualizing clinical workflows. Better understanding of clinical workflows can help care providers improve the quality, safety, and efficiency of providing care.

The paper describes each of these methods in detail.

The paper describes each of these methods in detail.

Tweet reply to comment

#FridayRoundsLive - Friday Rounds With Pritika Dasgupta

We discuss using accelerometers to identify people who are a risk of falling, why this "Data Queen" was drawn to biomedical informatics and key aspects of managing chronic conditions while in grad school.

Follow Pritika on twitter @pritikadasgupta.