Medical researchers are awash in a tsunami of clinical data. But we want major changes in how we gather, share, and apply this data to bring its advantages to all, says Leo Anthony Celi, principal research scientist on the MIT Laboratory for Computational Physiology (LCP). 

One key change is to make clinical data of all types openly available, with the correct privacy safeguards, says Celi, a practicing intensive care unit (ICU) physician on the Beth Israel Deaconess Medical Center (BIDMC) in Boston. One other key’s to totally exploit these open data with multidisciplinary collaborations amongst clinicians, academic investigators, and industry. A 3rd key’s to deal with the various needs of populations across every country, and to empower the experts there to drive advances in treatment, says Celi, who can also be an associate professor at Harvard Medical School. 

In all of this work, researchers must actively seek to beat the perennial problem of bias in understanding and applying medical knowledge. This deeply damaging problem is just heightened with the large onslaught of machine learning and other artificial intelligence technologies. “Computers will pick up all our unconscious, implicit biases after we make decisions,” Celi warns.

Sharing medical data 

Founded by the LCP, the MIT Critical Data consortium builds communities across disciplines to leverage the information which might be routinely collected within the means of ICU care to grasp health and disease higher. “We connect people and align incentives,” Celi says. “With the intention to advance, hospitals must work with universities, who must work with industry partners, who need access to clinicians and data.” 

The consortium’s flagship project is the MIMIC (medical information marked for intensive care) ICU database built at BIDMC. With about 35,000 users world wide, the MIMIC cohort is probably the most widely analyzed in critical care medicine. 

International collaborations resembling MIMIC highlight one in all the largest obstacles in health care: most clinical research is performed in wealthy countries, typically with most clinical trial participants being white males. “The findings of those trials are translated into treatment recommendations for each patient world wide,” says Celi. “We predict that this can be a major contributor to the sub-optimal outcomes that we see within the treatment of all styles of diseases in Africa, in Asia, in Latin America.” 

To repair this problem, “groups who’re disproportionately burdened by disease needs to be setting the research agenda,” Celi says. 

That is the rule within the “datathons” (health hackathons) that MIT Critical Data has organized in greater than two dozen countries, which apply the newest data science techniques to real-world health data. On the datathons, MIT students and school each learn from local experts and share their very own skill sets. Lots of these several-day events are sponsored by the MIT Industrial Liaison Program, the MIT International Science and Technology Initiatives program, or the MIT Sloan Latin America Office. 

Datathons are typically held in that country’s national language or dialect, moderately than English, with representation from academia, industry, government, and other stakeholders. Doctors, nurses, pharmacists, and social staff meet up with computer science, engineering, and humanities students to brainstorm and analyze potential solutions. “They need one another’s expertise to totally leverage and discover and validate the knowledge that’s encrypted in the information, and that will probably be translated into the way in which they deliver care,” says Celi. 

“In every single place we go, there’s incredible talent that is totally able to designing solutions to their health-care problems,” he emphasizes. The datathons aim to further empower the professionals and students within the host countries to drive medical research, innovation, and entrepreneurship.

Fighting built-in bias 

Applying machine learning and other advanced data science techniques to medical data reveals that “bias exists in the information in unimaginable ways” in every sort of health product, Celi says. Often this bias is rooted within the clinical trials required to approve medical devices and therapies. 

One dramatic example comes from pulse oximeters, which give readouts on oxygen levels in a patient’s blood. It seems that these devices overestimate oxygen levels for people of color. “Now we have been under-treating individuals of color since the nurses and the doctors have been falsely assured that their patients have adequate oxygenation,” he says. “We predict that we now have harmed, if not killed, lots of individuals prior to now, especially during Covid, in consequence of a technology that was not designed with inclusive test subjects.” 

Such dangers only increase because the universe of medical data expands. “The information that we now have available now for research is possibly two or three levels of magnitude greater than what we had even 10 years ago,” Celi says. MIMIC, for instance, now includes terabytes of X-ray, echocardiogram, and electrocardiogram data, all linked with related health records. Such enormous sets of information allow investigators to detect health patterns that were previously invisible. 

“But there’s a caveat,” Celi says. “It’s trivial for computers to learn sensitive attributes that should not very obvious to human experts.” In a study released last yr, as an illustration, he and his colleagues showed that algorithms can tell if a chest X-ray image belongs to a white patient or person of color, even without taking a look at some other clinical data. 

“More concerningly, groups including ours have demonstrated that computers can learn easily in case you’re wealthy or poor, just out of your imaging alone,” Celi says. “We were in a position to train a pc to predict in case you are on Medicaid, or if you will have private insurance, in case you feed them with chest X-rays with none abnormality. So again, computers are catching features that should not visible to the human eye.” And these features may lead algorithms to advise against therapies for people who find themselves Black or poor, he says. 

Opening up industry opportunities 

Every stakeholder stands to profit when pharmaceutical firms and other health-care corporations higher understand societal needs and might goal their treatments appropriately, Celi says. 

“We’d like to bring to the table the vendors of electronic health records and the medical device manufacturers, in addition to the pharmaceutical firms,” he explains. “They should be more aware of the disparities in the way in which that they perform their research. They should have more investigators representing underrepresented groups of individuals, to offer that lens to give you higher designs of health products.” 

Corporations may gain advantage by sharing results from their clinical trials, and will immediately see these potential advantages by participating in datathons, Celi says. “They might really witness the magic that happens when that data is curated and analyzed by students and clinicians with different backgrounds from different countries. So we’re calling out our partners within the pharmaceutical industry to prepare these events with us!”