Science & Enterprise logo
Science for business people. Enterprise for scientists.

AI Harnessed for Patient Monitoring Analytics

Comet screen display
Comet software screen display. Click on image for full-size view. (AMP3D Inc.)

25 Jan. 2021. New software is being trialed that captures vital signs and uses artificial intelligence to predict and display health trajectories for the patient. The software, called Continuous Monitoring of Event Trajectories, or Comet, is the work of researchers at University of Virginia in Charlottesville, and the spin-off company AMP3D, also in Charlottesville.

Comet aims to provide clinicians who monitor hospitalized patients with visual, real time displays of the direction of each patient’s health status. Most current systems provide a picture of a patient’s individual vital signs, such as heart rate or blood oxygen level, but clinicians need to combine those readings with lab test results and information from the patient’s health records to assess the person’s overall health status and future directions.

A UVa team led by medical school and biomedical engineering professor Randall Moorman seeks to provide nurses and doctors with better and faster tools to assess a patient’s condition. “Vital sign measurements and labs can come too late,” says Moorman in a university statement, “but early detection through predictive analytics has the power to improve patients’ outcomes, especially for catastrophic illnesses like Covid-19.”

Comet calculates and displays on a screen predictions of a patient’s cardiovascular and respiratory status. The data are provided by continuous monitoring of key heart and respiratory indicators, but also information from the patient’s health records, including latest lab results. Comet then applies algorithms to calculate future directions of the patient’s cardiovascular and respiratory status, and displays those predictive calculations.

Color-coded shooting stars

The visual display portrays predictions as icons shaped like shooting stars, with a large head and diminishing tail, aimed in a direction indicating improvement or deterioration of cardiovascular or respiratory stability, each measured on a separate axis. The shooting star icons are also color-coded beginning with yellow, then turning orange or red if the patient’s condition points to deterioration. Data from patients are later incorporated into the machine learning algorithm to refine its calculations.

UVa nursing school professor Jessica Keim Malpass and Liza Moorman of AMP3D describe Comet’s predictive analytics for Covid-19 patient monitoring in the International Journal of Nursing Studies Advances. In the paper, scheduled for the November 2021 issue, the authors note that Comet’s predictive analytics offer an early warning that patients are headed for trouble before symptoms develop. “Using precision predictive analytics systems like this one,” says Keim Malpass, “helps nurses initiate clinical response before the scenario becomes, quite literally, life and death.”

UVa is testing the Comet system in a clinical trial among cardiology patients at the university’s medical center. The study is enrolling 10,424 cardiac patients over two years, randomly assigned to receive standard care and be monitored by Comet with a display screen at bedside, or standard care alone. The study team is looking primarily for deterioration in patients’ conditions, such as need for transfer to an intensive care unit or life-threatening event such as cardiac arrest, within 21 days after admission.

Randall Moorman is also founder of AMP3D, short for Advanced Medical Predictive Devices, Diagnostics And Displays Inc., and serves as the company’s chief medical officer. AMP3D, which began in 2014, is commercializing the Comet software and extending its read-outs to mobile devices.

More from Science & Enterprise:

*     *     *