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.
At the “Apple Store of Doctor’s Offices,” Preventive Care Is High Tech. Is it Worth $150 a Month?
What if going to the doctor's office could be … nice?
If you didn't have to wait for your appointment, but were ushered right in; if your medical data was all collated and easily searchable on an iPhone app; if a remote scribe took notes while you spoke with your doctor so you could make eye contact with them; if your doctor didn't seem horribly rushed.
Would you go to the doctor to get help staying healthy, rather than just to stop being sick?
Would that change the way you thought about your health? Would you go to the doctor to get help staying healthy, rather than just to stop being sick? And would that, in the long run, be much better for you?
Those are the animating questions for Forward, a healthcare startup devoted to preventive care. Led by founder Adrian Aoun, formerly of Google/Sidewalk labs, Forward opened its first office in San Francisco in 2016 and has since expanded to Los Angeles, Orange County, New York, and Washington, D.C., with a San Diego location opening soon.
It's been described as the "Apple Store of doctor's offices," which in some ways is a reaction to Forward's vibe: Patients have described the offices as having blonde wood, minimalist design, sparkling water on tap — and lots of high-tech gadgets, like the full-body scanner that replaces the standard scale and stethoscope.
The interior of a Forward office.
(Courtesy Forward)
The more crucial difference, though, is its model of care. Forward doesn't take insurance. Instead, patients, or "members," pay a flat $149 per month, along the lines of a subscription service like Netflix or a gym membership. That fee covers visits, messaging with medical staff through the Forward app, the use of a wearable (like a Fitbit or a sleep tracker) if the physician recommends it, plus any bloodwork or diagnostic tests run in the on-site labs. (The company declined to disclose how many people have signed up for memberships.)
Predictability is Forward's other significant, distinguishing feature: No surprise co-pays, or extra charges showing up on a billing statement months later. Everything is wrapped up in the $149 membership fee, unless the physician recommends visiting an outside specialist.
That caveat isn't a small one. It's important to note that Forward is in no way meant to replace standard health insurance. The service is strictly focused on preventive care, so it wouldn't be much use in case of an emergency; it's meant to help people, as far as is possible, avoid that emergency at all.
Ani Okkasian's family recently went through such an emergency. Her 62-year-old father, an active and seemingly healthy man living with diabetes, had been feeling unwell for a while, but struggled to receive constructive follow-up or tests from his doctor. It finally emerged that his liver was severely damaged, and he suffered a stroke — the risk of which can be elevated by liver disease. He seemed to deteriorate completely within mere weeks, Okkasian said, and in January he passed away.
"He was someone who'd go to the doctor regularly and listen to what they said and follow it," Okkasian said. "I shouldn't have had to bury my father at 62. I still believe to my core that his death could have been avoided if the primary care was adequate."
"I could tell that the people who designed [Forward] had lost someone to the legacy system; it was so streamlined and so much clearer."
Okkasian began researching, looking for a better alternative, and discovered Forward. Founder Aoun lost his grandfather to a heart attack; his brother's heart attack at age 31 was the impetus to start Forward.
"I could tell that that was the genesis," Okkasian said. "Having just lost someone, and having had to deal with different aspects of the healthcare industry — how complicated and convoluted that all is — I could tell that the people who designed [Forward] had lost someone to the legacy system; it was so streamlined and so much clearer."
So Who Is Forward For?
The Affordable Care Act mandates that evidence-based preventive care must be covered by insurers without any cost to the patient. Today, 30 million Americans are still living without health insurance; but for most of the population, cost shouldn't prevent access to standard, preventive care, says Benjamin Sommers, a physician and professor at the Harvard T.H. Chan School of Public Health who has studied the effect of the ACA on preventive care access.
For Okkasian and her family, it wasn't a lack of access to primary care that was at issue; it was the quality of that primary care. In 2019, that's probably true for a lot of people.
"How come all other industries have been disturbed except the medical industry?" Okkasian asked. "It's disturbing the most people. We're so advanced in so many ways, but when it comes to the healthcare system, we're not prioritizing the wellness of a person."
Is Forward the answer? Well, probably not for everyone. Its office are only in a handful of cities, and there are limits to how scalable it would be; it's unavoidable that the $149 per month charge restricts access for a lot of people. Those who have insurance through their employer might have a flexible spending account (FSA) that would cover some or all of the membership fee, and Forward has said that 15 percent of their early members came from underserved communities and were offered free plans; but for many others, that's just an unworkable extra cost.
Sommers also sounded a dubious note about a maximalist attitude toward data collection.
"Even though some patients may think that 'more is always better' — more testing, more screening, etc. — this isn't true," he said. "Some types of cancer screening, ovarian cancer screening for instance, are actually harmful or of no benefit, because studies have shown that they don't improve survival or health outcomes, but can lead to unnecessary testing, pain, false positives, anxiety, and other side effects.
"It's really great for people who are in good health, looking to make it better."
"I'm generally skeptical of efforts to charge people more to get 'extra testing' that isn't currently supported by the medical evidence," he added.
But relatively healthy people who want to take a more active approach to their health — or people who have frequent testing needs, like those using the HIV-prevention drug PrEP, and want to avoid co-pays — might benefit from the on-demand, low-friction experience that Forward offers.
"It's really great for people who are in good health, looking to make it better," Okkasian said. "Your experience is simplified to a point where you feel empowered, not scared."
Facial Recognition Can Reduce Racial Profiling and False Arrests
[Editor's Note: This essay is in response to our current Big Question, which we posed to experts with different perspectives: "Do you think the use of facial recognition technology by the police or government should be banned? If so, why? If not, what limits, if any, should be placed on its use?"]
Opposing facial recognition technology has become an article of faith for civil libertarians. Many who supported the bans in cities like San Francisco and Oakland have declared the technology to be inherently racist and abusive.
The greatest danger would be to categorically oppose this technology and pretend that it will simply go away.
I have spent my career as a criminal defense attorney and a civil libertarian -- and I do not fear it. Indeed, I see it as positive so long as it is appropriately regulated and controlled.
We are living in the beginning of a biometric age, where technology uses our physical or biological characteristics for a variety of products and services. It holds great promises as well as great risks. The greatest danger, however, would be to categorically oppose this technology and pretend that it will simply go away.
This is an age driven as much by consumer as it is government demand. Living in denial may be emotionally appealing, but it will only hasten the creation of post-privacy world. If we do not address this emerging technology, movements in public will increasingly result in instant recognition and even tracking. It is the type of fish-bowl society that strips away any expectation of privacy in our interactions and associations.
The biometrics field is expanding exponentially, largely due to the popularity of consumer products using facial recognition technology (FRT) -- from the iPhone program to shopping ones that recognize customers.
But the privacy community is losing this battle because it is using the privacy rationales and doctrines forged in the earlier electronic surveillance periods. Just as generals are often accused of planning to fight the last war, civil libertarians can sometimes cling to past models despite their decreasing relevance in the current world.
I see FRT as having positive implications that are worth pursuing. When properly used, biometrics can actually enhance privacy interests and even reduce racial profiling by reducing false arrests and the warrantless "patdowns" allowed by the Supreme Court. Bans not only deny police a technology widely used by businesses, but return police to the highly flawed default of "eye balling" suspects -- a system with a considerably higher error rate than top FRT programs.
Officers are often wrong and stop a great number of suspects in the hopes of finding a wanted felon.
A study in Australia showed that passport officers who had taken photographs of subjects in ideal conditions nonetheless experienced high error rates when identifying them shortly afterward, including 14 percent false acceptance rates. Currently, officers stop suspects based on their memory from seeing a photograph days or weeks earlier. They are often wrong and stop a great number of suspects in the hopes of finding a wanted felon. The best FRT programs achieve an astonishing accuracy rate, though real-world implementation has challenges that must be addressed.
One legitimate concern raised in early studies showed higher error rates in recognitions for certain groups, particularly African American women. An MIT study finding that error rate prompted major improvements in the algorithms as well as training changes to greatly reduce the frequency of errors. The issue remains a concern, but there is nothing inherently racist in algorithms. These are a set of computer instructions that isolate and process with the parameters and conditions set by creators.
To be sure, there is room for improvement in some algorithms. Tests performed by the American Civil Liberties Union (ACLU) reportedly showed only an 80 percent accuracy rate in comparing mug shots to pictures of members of Congress when using Amazon's "Rekognition" system. It recently showed the same 80 percent rate in doing the same comparison to members of the California legislators.
However, different algorithms are available with differing levels of performance. Moreover, these products can be set with a lower discrimination level. The fact is that the top algorithms tested by the National Institute of Standards and Technology showed that their accuracy rate is greater than 99 percent.
The greatest threat of biometric technologies is to democratic values.
Assuming a top-performing algorithm is used, the result could be highly beneficial for civil liberties as opposed to the alternative of "eye balling" suspects. Consider the Boston Bombing where police declared a "containment zone" and forced families into the street with their hands in the air.
The suspect, Dzhokhar Tsarnaev, moved around Boston and was ultimately found outside the "containment zone" once authorities abandoned near martial law. He was caught on some surveillance systems but not identified. FRT can help law enforcement avoid time-consuming area searches and the questionable practice of forcing people out of their homes to physically examine them.
If we are to avoid a post-privacy world, we will have to redefine what we are trying to protect and reconceive how we hope to protect it. In my view, the greatest threat of biometric technologies is to democratic values. Authoritarian nations like China have made huge investments into FRT precisely because they know that the threat of recognition in public deters citizens from associating or interacting with protesters or dissidents. Recognition changes conduct. That chilling effect is what we have the worry about the most.
Conventional privacy doctrines do not offer much protection. The very concept of "public privacy" is treated as something of an oxymoron by courts. Public acts and associations are treated as lacking any reasonable expectation of privacy. In the same vein, the right to anonymity is not a strong avenue for protection. We are not living in an anonymous world anymore.
Consumers want products like FaceFind, which link their images with others across social media. They like "frictionless" transactions and authentications using faceprints. Despite the hyperbole in places like San Francisco, civil libertarians will not succeed in getting that cat to walk backwards.
The basis for biometric privacy protection should not be focused on anonymity, but rather obscurity. You will be increasingly subject to transparency-forcing technology, but we can legislatively mandate ways of obscuring that information. That is the objective of the Biometric Privacy Act that I have proposed in recent research. However, no such comprehensive legislation has passed through Congress.
The ability to spot fraudulent entries at airports or recognizing a felon in flight has obvious benefits for all citizens.
We also need to recognize that FRT has many beneficial uses. Biometric guns can reduce accidents and criminals' conduct. New authentications using FRT and other biometric programs could reduce identity theft.
And, yes, FRT could help protect against unnecessary police stops or false arrests. Finally, and not insignificantly, this technology could stop serious crimes, from terrorist attacks to the capturing of dangerous felons. The ability to spot fraudulent entries at airports or recognizing a felon in flight has obvious benefits for all citizens.
We can live and thrive in a biometric era. However, we will need to bring together civil libertarians with business and government experts if we are going to control this technology rather than have it control us.
[Editor's Note: Read the opposite perspective here.]