How a bad day at work led to better COVID predictions

Credit: Unsplash / CC0 Public Domain

Talking about your bad day at work can lead to great solutions. Cold Spring Harbor Laboratory (CSHL) Associate Professor Saket Navlakha and his wife, Dr. Sejal Morjaria, an infectious disease doctor at Memorial Sloan Kettering Cancer Center (MSK), found a way to predict the severity of COVID-19 in cancer patients. The calculation tool they have developed prevents unnecessarily expensive tests and improves patient care.

Morjaria says, “In general, I have a good intuition about how patients will develop.” However, that intuition failed her when confronted with COVID-19. She says:

“When the pandemic first emerged, we found it difficult to understand and predict which patients would have severe COVID. People ordered a number of laboratories and there were often unnecessary laboratory tests.”

Navlakha joined CSHL in 2019. He uses computer science to understand biological processes. Morjaria wondered if her husband could help:

“So I came home and said, ‘Saket, it would be great if we could find a way to use machine learning to find out which patients will or will not develop severe COVID later.’ “”

The team collected 267 variables from cancer patients diagnosed with COVID-19. Variables ranged from age and gender to cancer type, recent treatments, and laboratory results. They trained a machine-learning computer program to divide patients into three groups. Those who need a lot of oxygen from a ventilator:

  1. right away
  2. after a few days
  3. not at all

The researchers found around 50 variables that contributed the most to predicting outcomes. Their method had an accuracy rate of 70-85% and was particularly suitable for patients who would need immediate ventilation. In general, the tool can help distinguish interactions between multiple risk factors that may not be noticeable even to those with trained eyes. The program also prevents overtests, which Morjaria knows will “save patients unnecessary massive hospital costs”.

Navlakha believes this work would not have been possible without working closely with his wife and other MSK scientists, including Rocio-Perez Johnston and Ying Taur. He says:

“Sejal and I are talking about better ways to integrate what she experiences at the bedside than what we can analyze and do computationally. As someone who has never worked with clinical data, I would have tried to do this without Sejal’s guide to doing tons of mistakes, it would just have been a total disaster and completely useless. “

Navlakha and Morjaria hope their work will inspire more doctors and computer scientists to work together and develop innovative clinical solutions for complex diseases.

British cancer patients are more likely to die after COVID-19 than European cancer patients

More information:
BMC Infectious Diseases, DOI: 10.1186 / s12879-021-06038-2 Provided by the Cold Spring Harbor Laboratory

Quote: How a Bad Work Day Led to Better COVID Predictions (2021, May 3), accessed May 3, 2021 from

This document is subject to copyright. Except for fair trade for the purpose of private study or research, no part may be reproduced without written permission. The content is provided for informational purposes only.