Scientists and doctors tell the NBC Bay Area Investigative Unit the field of data crunching, often referred to as “big data,” could one day reveal who is most vulnerable to the deadliest strains of COVID-19. The process of analyzing patient health records to optimize treatments is already being credited for freeing up valuable space in intensive care units by arming physicians with much-needed information to discharge COVID-19 patients at a much faster pace.
While some physicians welcome such a data tool in their hospitals, others worry the nation’s medical system may not be able to take full advantage so long as individual health centers remain reluctant to opening their own patient data vaults to competitors.
Data Could Prove to be Valuable Medicine Against COVID-19
Medical workers remain desperate to find an edge with each spike of coronavirus cases. Doctors and nurses worry about running out of hospital beds, especially in the Intensive Care Unit, where the most serious cases of COVID-19 are treated. A team of data scientists at Stanford, however, has found a way to boost ICU capacity by as much as 25%.
“I would have imagined that people who were sent home with oxygen would have a higher chance of getting complications. That did not turn out to be the case,” said Dr. Nigam Shah, a data scientist and a professor of Medical Bioinformatics at Stanford University.
This is like getting a second opinion
Dr. Nigam Shah, professor of Medical Bioinformatics at Stanford University
By analyzing health records of past COVID-19 patients at Stanford Hospital, Dr. Shah and his team discovered that in many cases, patients released from ICU and sent home did just as well as those who remained in the ICU.
Local
“There was practically no difference,” said Dr. Shah. “To me, it was surprising.”
Shah and his team are building a database of COVID-19 patients that includes the race, age, medical history, and the course of treatment that worked and didn’t work. “This is like getting a second opinion,” he said.
Such a database could ultimately help pinpoint who is at greater risk of serious illness from COVID-19 or other diseases. Doctors also hope to eventually rely on the data to help make personalized treatment decisions. As Dr. Shah points out, doctors frequently make treatment decisions based on the experiences of past patients, but a physician’s memory can be limited. Computer data, however, can provide an almost supernatural perspective.
“I can now see millions of patients and I can answer questions that no human doctor would be in a position to answer because it's just humanly not possible to hold the collective experience of a million patients in your head,” said Shah, who's been analyzing data for 15 years, but only recently began focusing on COVID-19.
How do we use the data that we have to inform the best practice?
Dr. Kevin Schulman, professor of medicine at Stanford University
The idea to study COVID-19 cases came from Dr. Kevin Schulman, who was looking for more resources in treating coronavirus patients at Stanford.
“How do we use the data that we have to inform the best practice," wondered Schulman.
Stanford Database Holds Records for 700 Coronavirus Patients
Roughly 700 COVID-19 patients are now included in Stanford’s database. While that’s not a big enough sample to rely on when making medical decisions for specific cases, Stanford is utilizing the data to draw some conclusions that are helping free up valuable space in the intensive care unit and get hospital patients home sooner.
Last spring and summer, Stanford hospital guidelines dictated that when patients needed 6 liters of oxygen, they were transferred to the ICU. But the data now confirms that in many cases, those patients will do just as well without the ICU.
“The more we can free up capacity in the intensive care units,” said Dr. Schulman, “The more opportunity we have to make sure we're ready to do additional cases [as they] come in.”
The statistics also showed patients who were sent home on portable oxygen systems did just as well as those who were plugged into ventilators in the ICU. Stanford doctors suspected this was true but needed the data to back it up. Armed with the data, Stanford changed its policies hospital-wide. If the average patient remains in Intensive Care for four days, eliminating one day can boost capacity by 25%.
“The quicker I can get them home safely, the better off they're going to be in the long run,” said Dr. Schulman, adding, “It's easier to recover at home with friends and family around.”
Assembling Patient Database May be Tough Pill to Swallow
Dr. Peter Chin-Hong, an infectious diseases expert and professor of medicine at UCSF, said he sees the potential for using a large patient data base but also anticipates major obstacles.
"Maybe when the dust settles and there's a steady state, I might look at that algorithm,” he said. "But right now, I'm just trying to put out the fire.”
Dr. Chin-Hong points out that thousands if not millions of patient cases would be needed for the system to be truly effective at predicting the best possible treatment. Additionally, competing hospital systems may not readily share their data.
“The more systems you bring together, the branding kind of gets murky ... so who pays for it?” said Dr. Chin-Hong. “It would take a lot of strong people going to the table with reassurances that everybody has a slice of the pie.”
Dr. Shah, however, says that kind of nationwide or even statewide partnership is the very kind of prescription COVID-19 patients need.
“Pretty much every other industry that I can think of uses information on what have they done in the past to make their services better,” he said. “COVID-19 has accelerated a decade worth of technological progress in one year, and if people buy into this idea that we can share data for benefiting the care of other patients, I think that's very powerful.”
