GE Healthcare and Oxford University seek to improve COVID-19 management and outcomes
There is so much we still don’t know about COVID-19. Why do some people get so sick and others don’t even show symptoms? Why is the virus so rare in children and teenagers? Why are people with underlying medical conditions six times more likely to be hospitalised and 12 times more likely to die? Which patients will suffer long-term consequences from COVID-related lung damage? And why are rates so much higher in the BAME population? There are even unknowns about how the virus is transmitted.
Now GE Healthcare and the University of Oxford-led National Consortium of Intelligent Medical Imaging (NCIMI) plan a study using artificial intelligence (AI) to try and answer some of these important questions. Together, the partners plan to develop and test algorithms to aid in the diagnosis and management of COVID-19.
“It would be extremely valuable to predict at a relatively early stage in the disease which patients will do well, which are at risk of imminent deterioration and should be admitted to ICU, and which are at higher risk of delayed deterioration and need to be actively monitored,” said Professor Fergus Gleeson, a Consultant Radiologist and Professor of Radiology at Oxford, who is leading the study. “These distinctions would allow hospital resources to be targeted to those who will need them while in hospital and following discharge.”
The planned UK-based AI-enhanced COVID-19 Prognostic Algorithm (HOST) trial proposes to focus on developing, enhancing, and testing potential algorithms to help diagnose COVID-19 pneumonia and predict which patients will experience severe respiratory distress—a key cause of mortality—and which might develop longer-term lung function problems, even when they recover from respiratory distress.
“If we can ensure patients are quickly placed in the right care setting, this may help improve outcomes,” said GE Healthcare President and CEO Kieran Murphy. This, in turn, could save valuable hospital and ICU beds for patients who need them most – even before their condition turns dire – while enabling lower-risk patients to ambulate at home.
GE Healthcare plans to use data from NCIMI partner hospitals and thousands of medical imaging, laboratory, and clinical observations from COVID patients, as well as the National COVID-19 Chest Imaging Database, to develop the imaging and vital sign algorithms, while the Oxford team would assess and test the algorithms and their potential role in the clinical setting. The goal is to develop tools that help manage COVID-19 patients from triage to acute monitoring, interventions, discharge, and short- and long-term follow up.
“We decided that if we could work with longitudinal data from patients, we could use AI to explore the links between COVID symptoms, biomarkers, patient risk factors, and outcomes,” said Ben Newton, General Manager, Oncology at GE Healthcare.. It’s an approach GE Healthcare has already developed for cancer, he said, including using algorithms to predict response to therapy. “This is re-purposing that same machinery in the organisation to work on this acute problem for COVID.”
Artificial Intelligence and COVID-19: A Perfect Match
GE Healthcare is pleased to be involved in these important research efforts surrounding AI and COVID-19, which Wired magazine predicts will accelerate the use of AI in healthcare. Indeed, it was an AI algorithm that first detected an unusual cluster of respiratory illnesses in Wuhan, China, the epicentre of the pandemic, several days before the World Health Organisation announced the outbreak.
There are already signs that AI is playing a major role in efforts to predict, diagnose, and treat COVID-19. A review of how AI is already being used in COVID identified seven key areas:
- Early detection and diagnosis
- Monitoring treatment
- Contact tracing
- Projecting cases and mortality
- Developing drugs and vaccines
- Reducing health care worker workload
- Identifying ways to prevent the disease
Researchers are also using AI to predict the risk of sepsis in COVID patients; design proteins to block the virus; and model the risk of infection as states and countries begin to open up.
The HOST trial, said Newton, “will take the deepest learning of past several months and aim to prevent deaths.”
 Beachum L, et al. Coronavirus has been 12 times more deadly for people with underlying conditions, CDC says. Washington Post. June 17, 2020. Available at: https://www.washingtonpost.com/nation/2020/06/15/coronavirus-live-updates-us/.
  Lee K. Covid-19 Will Accelerate the AI Health Care Revolution. Wired. May 22, 2020. Available at: https://www.wired.com/story/covid-19-will-accelerate-ai-health-care-revolution/.
 Artificial Intelligence (AI) applications for COVID-19 pandemic. Diab Metabolic Syndrome: Clinical Research & Reviews. 2020; 14(4): 337-339.
 Martineau K. Marshaling artificial intelligence in the fight against Covid-19. MIT News. May 19, 2020. Available at: http://news.mit.edu/2020/mit-marshaling-artificial-intelligence-fight-against-covid-19-0519