DBDS Student Talks (BIOMEDIN 201) Autumn Quarter 2024-25 

11/18/24

12:15-12:45: Bryan Bunning

12:45-1:15: Yixing Jiang

LK 120/or Zoom Link: https://stanford.zoom.us/j/97747187137?pwd=QJqAPX1yRpGOwTTGPsIIwEYp7W2Aaz.1&from=addon
Password: 180750

Title: A Micro-Randomized Trial Design of Remote Patient Monitoring in Pediatric Type 1 Diabetes

Abstract: At Stanford Children’s Hospital, the standard-of-care for Type 1 Diabetes (T1D) in children now includes continuous glucose monitors (CGMs) and AI-enabled remote patient monitoring (RPM) to track blood sugar levels in real time. This initiative, known as the Teamwork, Targets, Technology, and Tight Control (4T) Program, aggregates patient glucose data, sends it to the cloud for analysis, and presents it to the care team via a dashboard to support timely insulin adjustments and patient triage.

In this study, we designed a micro-randomized trial embedded within the 4T Program for newly diagnosed pediatric T1D patients. The intervention involves increasing the RPM frequency for a subpopulation of patients who are not meeting their glucose targets. Through simulations, we assessed the impact of factors such as study duration, sample size, and clinic capacity. Clinic capacity, ie the working capacity for the staff to act and provide care from the remote data, is highlighted. This simulation-based approach provides a practical framework for designing effective RPM studies under real-world constraints. This trial may offer evidence to inform RPM billing policy standards in diabetes.

Yixing Jiang, 3rd Year PhD Student

(12:45pm-1:15pm)

Title: SmartAlert: Integrating Machine Learning and Alert Triggers into Live Electronic Medical Record Systems, Targeting Low-Yield Inpatient Lab Tests

Abstract: This study explores integrating machine learning into electronic medical record systems to predict stability of inpatient lab tests. A ‘smart alerts’ system was developed and tested at Stanford Hospital. The system identifies stable lab results, advising clinicians on test ordering. Live deployment showed desired precision at good recall in predicting test result stability, with suggestions for system optimization identified. This approach may significantly decrease low-yield testing and enhance personalized clinical decision-making.