Artificial Intelligence is getting better than humans at detecting breast cancer
Since the early 2000s, AI systems have eliminated more than 1.7 million jobs, and that number will only increase as AI improves. Some research estimates that by 2025, AI will eliminate more than 85 million jobs.
But for all the talk about job security, AI is also proving to be a powerful tool in healthcare—specifically, cancer detection. One recently published study has shown that, remarkably, artificial intelligence was able to detect 20 percent more cancers in imaging scans than radiologists alone.
Published in The Lancet Oncology, the study analyzed the scans of 80,000 Swedish women with a moderate hereditary risk of breast cancer who had undergone a mammogram between April 2021 and July 2022. Half of these scans were read by AI and then a radiologist to double-check the findings. The second group of scans was read by two researchers without the help of AI. (Currently, the standard of care across Europe is to have two radiologists analyze a scan before diagnosing a patient with breast cancer.)
The study showed that the AI group detected cancer in 6 out of every 1,000 scans, while the radiologists detected cancer in 5 per 1,000 scans. In other words, AI found 20 percent more cancers than the highly-trained radiologists.
Scientists have been using MRI images (like the ones pictured here) to train artificial intelligence to detect cancers earlier and with more accuracy. Here, MIT's AI system, MIRAI, looks for patterns in a patient's mammograms to detect breast cancer earlier than ever before. news.mit.edu
But even though the AI was better able to pinpoint cancer on an image, it doesn’t mean radiologists will soon be out of a job. Dr. Laura Heacock, a breast radiologist at NYU, said in an interview with CNN that radiologists do much more than simply screening mammograms, and that even well-trained technology can make errors. “These tools work best when paired with highly-trained radiologists who make the final call on your mammogram. Think of it as a tool like a stethoscope for a cardiologist.”
AI is still an emerging technology, but more and more doctors are using them to detect different cancers. For example, researchers at MIT have developed a program called MIRAI, which looks at patterns in patient mammograms across a series of scans and uses an algorithm to model a patient's risk of developing breast cancer over time. The program was "trained" with more than 200,000 breast imaging scans from Massachusetts General Hospital and has been tested on over 100,000 women in different hospitals across the world. According to MIT, MIRAI "has been shown to be more accurate in predicting the risk for developing breast cancer in the short term (over a 3-year period) compared to traditional tools." It has also been able to detect breast cancer up to five years before a patient receives a diagnosis.
The challenges for cancer-detecting AI tools now is not just accuracy. AI tools are also being challenged to perform consistently well across different ages, races, and breast density profiles, particularly given the increased risks that different women face. For example, Black women are 42 percent more likely than white women to die from breast cancer, despite having nearly the same rates of breast cancer as white women. Recently, an FDA-approved AI device for screening breast cancer has come under fire for wrongly detecting cancer in Black patients significantly more often than white patients.
As AI technology improves, radiologists will be able to accurately scan a more diverse set of patients at a larger volume than ever before, potentially saving more lives than ever.
Antibody Testing Alone is Not the Key to Re-Opening Society
[Editor's Note: We asked experts from different specialties to weigh in on a timely Big Question: "How should immunity testing play a role in re-opening society?" Below, a virologist offers her perspective.]
With the advent of serology testing and increased emphasis on "re-opening" America, public health officials have begun considering whether or not people who have recovered from COVID-19 can safely re-enter the workplace.
"Immunity certificates cannot certify what is not known."
Conventional wisdom holds that people who have developed antibodies in response to infection with SARS-CoV-2, the coronavirus that causes COVID-19, are likely to be immune to reinfection.
For most acute viral infections, this is generally true. However, SARS-CoV-2 is a new pathogen, and there are currently many unanswered questions about immunity. Can recovered patients be reinfected or transmit the virus? Does symptom severity determine how protective responses will be after recovery? How long will protection last? Understanding these basic features is essential to phased re-opening of the government and economy for people who have recovered from COVID-19.
One mechanism that has been considered is issuing "immunity certificates" to individuals with antibodies against SARS-CoV-2. These certificates would verify that individuals have already recovered from COVID-19, and thus have antibodies in their blood that will protect them against reinfection, enabling them to safely return to work and participate in society. Although this sounds reasonable in theory, there are many practical reasons why this is not a wise policy decision to ease off restrictive stay-home orders and distancing practices.
Too Many Scientific Unknowns
Serology tests measure antibodies in the serum—the liquid component of blood, which is where the antibodies are located. In this case, serology tests measure antibodies that specifically bind to SARS-CoV-2 virus particles. Usually when a person is infected with a virus, they develop antibodies that can "recognize" that virus, so the presence of SARS-CoV-2 antibodies indicates that a person has been previously exposed to the virus. Broad serology testing is critical to knowing how many people have been infected with SARS-CoV-2, since testing capacity for the virus itself has been so low.
Tests for the virus measure amounts of SARS-CoV-2 RNA—the virus's genetic material—directly, and thus will not detect the virus once a person has recovered. Thus, the majority of people who were not severely ill and did not require hospitalization, or did not have direct contact with a confirmed case, will not test positive for the virus weeks after they have recovered and can only determine if they had COVID-19 by testing for antibodies.
In most cases, for most pathogens, antibodies are also neutralizing, meaning they bind to the virus and render it incapable of infecting cells, and this protects against future infections. Immunity certificates are based on the assumption that people with antibodies specific for SARS-CoV-2 will be protected against reinfection. The problem is that we've only known that SARS-CoV-2 existed for a little over four months. Although studies so far indicate that most (but not all) patients with confirmed COVID-19 cases develop antibodies, we don't know the extent to which antibodies are protective against reinfection, or how long that protection will last. Immunity certificates cannot certify what is not known.
The limited data so far is encouraging with regard to protective immunity. Most of the patient sera tested for antibodies show reasonable titers of IgG, the type of antibodies most likely to be neutralizing. Furthermore, studies have shown that these IgG antibodies are capable of neutralizing surrogate viruses as well as infectious SARS-CoV-2 in laboratory tests. In addition, rhesus monkeys that were experimentally infected with SARS-CoV-2 and allowed to recover were protected from reinfection after a subsequent experimental challenge. These data tentatively suggest that most people are likely to develop neutralizing IgG, and protective immunity, after being infected by SARS-CoV-2.
However, not all COVID-19 patients do produce high levels of antibodies specific for SARS-CoV-2. A small number of patients in one study had no detectable neutralizing IgG. There have also been reports of patients in South Korea testing PCR positive after a prior negative test, indicating reinfection or reactivation. These cases may be explained by the sensitivity of the PCR test, and no data have been produced to indicate that these cases are genuine reinfection or recurrence of viral infection.
Complicating matters further, not all serology tests measure antibody titers. Some rapid serology tests are designed to be binary—the test can either detect antibodies or not, but does not give information about the amount of antibodies circulating. Based on our current knowledge, we cannot be certain that merely having any level of detectable antibodies alone guarantees protection from reinfection, or from a subclinical reinfection that might not cause a second case of COVID-19, but could still result in transmission to others. These unknowns remain problematic even with tests that accurately detect the presence of antibodies—which is not a given today, as many of the newly available tests are reportedly unreliable.
A Logistical and Ethical Quagmire
While most people are eager to cast off the isolation of physical distancing and resume their normal lives, mere desire to return to normality is not an indicator of whether those antibodies actually work, and no certificate can confer immune protection. Furthermore, immunity certificates could lead to some complicated logistical and ethical issues. If antibodies do not guarantee protective immunity, certifying that they do could give antibody-positive people a false sense of security, causing them to relax infection control practices such as distancing and hand hygiene.
"We should not, however, place our faith in assumptions and make return to normality contingent on an arbitrary and uninformative piece of paper."
Certificates could be forged, putting susceptible people at higher exposure risk. It's not clear who would issue them, what they would entitle the bearer to do or not do, or how certification would be verified or enforced. There are many ways in which such certificates could be used as a pretext to discriminate against people based on health status, in addition to disability, race, and socioeconomic status. Tracking people based on immune status raises further concerns about privacy and civil rights.
Rather than issuing documents confirming immune status, we should instead "re-open" society cautiously, with widespread virus and serology testing to accurately identify and isolate infected cases rapidly, with immediate contact tracing to safely quarantine and monitor those at exposure risk. Broad serosurveillance must be coupled with functional assays for neutralization activity to begin assessing how protective antibodies might actually be against SARS-CoV-2 infection. To understand how long immunity lasts, we should study antibodies, as well as the functional capabilities of other components of the larger immune system, such as T cells, in recovered COVID-19 patients over time.
We should not, however, place our faith in assumptions and make return to normality contingent on an arbitrary and uninformative piece of paper. Re-opening society, the government, and the economy depends not only on accurately determining how many people have antibodies to SARS-CoV-2, but on a deeper understanding of how those antibodies work to provide protection.
Harvard Researchers Are Using a Breakthrough Tool to Find the Antibodies That Best Knock Out the Coronavirus
To find a cure for a deadly infectious disease in the 1995 medical thriller Outbreak, scientists extract the virus's antibodies from its original host—an African monkey.
"When a person is infected, the immune system makes antibodies kind of blindly."
The antibodies prevent the monkeys from getting sick, so doctors use these antibodies to make the therapeutic serum for humans. With SARS-CoV-2, the original hosts might be bats or pangolins, but scientists don't have access to either, so they are turning to the humans who beat the virus.
Patients who recovered from COVID-19 are valuable reservoirs of viral antibodies and may help scientists develop efficient therapeutics, says Stephen J. Elledge, professor of genetics and medicine at Harvard Medical School in Boston. Studying the structure of the antibodies floating in their blood can help understand what their immune systems did right to kill the pathogen.
When viruses invade the body, the immune system builds antibodies against them. The antibodies work like Velcro strips—they use special spots on their surface called paratopes to cling to the specific spots on the viral shell called epitopes. Once the antibodies circulating in the blood find their "match," they cling on to the virus and deactivate it.
But that process is far from simple. The epitopes and paratopes are built of various peptides that have complex shapes, are folded in specific ways, and may carry an electrical charge that repels certain molecules. Only when all of these parameters match, an antibody can get close enough to a viral particle—and shut it out.
So the immune system forges many different antibodies with varied parameters in hopes that some will work. "When a person is infected, the immune system makes antibodies kind of blindly," Elledge says. "It's doing a shotgun approach. It's not sure which ones will work, but it knows once it's made a good one that works."
Elledge and his team want to take the guessing out of the process. They are using their home-built tool VirScan to comb through the blood samples of the recovered COVID-19 patients to see what parameters the efficient antibodies should have. First developed in 2015, the VirScan has a library of epitopes found on the shells of viruses known to afflict humans, akin to a database of criminals' mug shots maintained by the police.
Originally, VirScan was meant to reveal which pathogens a person overcame throughout a lifetime, and could identify over 1,000 different strains of viruses and bacteria. When the team ran blood samples against the VirScan's library, the tool would pick out all the "usual suspects." And unlike traditional blood tests called ELISA, which can only detect one pathogen at a time, VirScan can detect all of them at once. Now, the team has updated VirScan with the SARS-CoV-2 "mug shot" and is beginning to test which antibodies from the recovered patients' blood will bind to them.
Knowing which antibodies bind best can also help fine-tune vaccines.
Obtaining blood samples was a challenge that caused some delays. "So far most of the recovered patients have been in China and those samples are hard to get," Elledge says. It also takes a person five to 10 days to develop antibodies, so the blood must be drawn at the right time during the illness. If a person is asymptomatic, it's hard to pinpoint the right moment. "We just got a couple of blood samples so we are testing now," he said. The team hopes to get some results very soon.
Elucidating the structure of efficient antibodies can help create therapeutics for COVID-19. "VirScan is a powerful technology to study antibody responses," says Harvard Medical School professor Dan Barouch, who also directs the Center for Virology and Vaccine Research. "A detailed understanding of the antibody responses to COVID-19 will help guide the design of next-generation vaccines and therapeutics."
For example, scientists can synthesize antibodies to specs and give them to patients as medicine. Once vaccines are designed, medics can use VirScan to see if those vaccinated again COVID-19 generate the necessary antibodies.
Knowing which antibodies bind best can also help fine-tune vaccines. Sometimes, viruses cause the immune system to generate antibodies that don't deactivate it. "We think the virus is trying to confuse the immune system; it is its business plan," Elledge says—so those unhelpful antibodies shouldn't be included in vaccines.
More importantly, VirScan can also tell which people have developed immunity to SARS-CoV-2 and can return to their workplaces and businesses, which is crucial to restoring the economy. Knowing one's immunity status is especially important for doctors working on the frontlines, Elledge notes. "The resistant ones can intubate the sick."
Lina Zeldovich has written about science, medicine and technology for Popular Science, Smithsonian, National Geographic, Scientific American, Reader’s Digest, the New York Times and other major national and international publications. A Columbia J-School alumna, she has won several awards for her stories, including the ASJA Crisis Coverage Award for Covid reporting, and has been a contributing editor at Nautilus Magazine. In 2021, Zeldovich released her first book, The Other Dark Matter, published by the University of Chicago Press, about the science and business of turning waste into wealth and health. You can find her on http://linazeldovich.com/ and @linazeldovich.