Columbia Science Review
  • Home
  • About
    • Executive Board
    • Editorial Board
  • Blog
  • Events
    • 2022-2023
    • 2021-2022
    • 2020-2021
    • 2019-2020
    • 2018-2019
    • 2017-2018
    • 2016-2017
  • Publications
  • COVID-19 Public Hub
    • Interviews >
      • Biology of COVID-19
      • Public Health
      • Technology & Data
    • Frontline Stories >
      • Healthcare Workers
      • Global Health
      • Volunteer Efforts
    • Resources & Links >
      • FAQ's
      • Resource Hubs
      • Student Opportunities
      • Podcasts & Graphics
      • Mental Health Resources
      • Twitter Feeds
      • BLM Resources
    • Columbia Events >
      • Campus Events
      • CUMC COVID-19 Symposium
      • CSR Events
    • Our Team
  • Contact

Sarah Rossetti and Kenrick Cato: Scaling up for Surge Capacity and COVID-19 Patient tracking in the EHR: Leveraging Healthcare process modeling (covid-19 symposium)

4/1/2020

0 Comments

 
Summary by: Arooba Ahmed (CC '23)

     Drs. Rossetti and Cato developed a model to help deal with the surge capacity, or the high influx of patients. The bulk of the work is done with the CONCERN (Communicating Narrative Concerns Entered by RNs) project. It is built off of the need for the hospital to dynamically classify patients to their beds. People manually enter information when updating beds, but when things are now changing frequently, validating the manual process becomes critical. 
     The data they present are US cases. The hospital system as a whole needs to understand foundational mapping, such as which beds existed before the surge. They also need to be synced with the ADT (Admission discharge and transfer system) and classified by adult or pediatric, care level, and care type (hallway beds etc). They also need to map whether they are physical or virtual beds. Virtual beds are locations where someone will be in the future, such a bed after leaving the emergency room. It is also important to identify cases which have been moved from one type of bed to another. They also wish to track ventilators.
     Using the healthcare processing model, they see how clinicians interact with the system to identify types of bed. This uses patterns of documentation and machine learning. They apply and validate these models with clinicians to see if the learning makes sense.  This model is based on patterns of actual clinician behavior, as it incorporates the person's clinical expertise. The method also incorporates information from DHA entries, which is where nurses put their own observations. 
     The model uses a multi class logistic regression so the data can be turned into a report and run at 24 hour intervals. It is quite precise overall, with the only issues being in areas like the obstetrics, where there are subclasses within each category. The model was designed to be scalable and so it can be used in a time like this.
0 Comments

Your comment will be posted after it is approved.


Leave a Reply.

Columbia Science Review
© COPYRIGHT 2022. ALL RIGHTS RESERVED.
  • Home
  • About
    • Executive Board
    • Editorial Board
  • Blog
  • Events
    • 2022-2023
    • 2021-2022
    • 2020-2021
    • 2019-2020
    • 2018-2019
    • 2017-2018
    • 2016-2017
  • Publications
  • COVID-19 Public Hub
    • Interviews >
      • Biology of COVID-19
      • Public Health
      • Technology & Data
    • Frontline Stories >
      • Healthcare Workers
      • Global Health
      • Volunteer Efforts
    • Resources & Links >
      • FAQ's
      • Resource Hubs
      • Student Opportunities
      • Podcasts & Graphics
      • Mental Health Resources
      • Twitter Feeds
      • BLM Resources
    • Columbia Events >
      • Campus Events
      • CUMC COVID-19 Symposium
      • CSR Events
    • Our Team
  • Contact