Artificial Intelligence is getting better than humans at detecting breast cancer
A recent study in The Lancet Oncology showed that AI found 20 percent more cancers on mammogram screens than radiologists alone.
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.
Short Story Contest Winner: "The Gerry Program"
A father and son gaze at the ducks in a lake.
It's an odd sensation knowing you're going to die, but it was a feeling Gerry Ferguson had become relatively acquainted with over the past two years. What most perplexed the terminally ill, he observed, was not the concept of death so much as the continuation of all other life.
Gerry's secret project had been in the works for two years now, ever since they found the growth.
Who will mourn me when I'm gone? What trait or idiosyncrasy will people most recall? Will I still be talked of, 100 years from now?
But Gerry didn't worry about these questions. He was comfortable that his legacy would live on, in one form or another. From his cozy flat in the west end of Glasgow, Gerry had managed to put his affairs in order and still find time for small joys.
Feeding the geese in summer at the park just down from his house, reading classics from the teeming bookcase in the living room, talking with his son Michael on Skype. It was Michael who had first suggested reading some of the new works of non-fiction that now littered the large oak desk in Gerry's study.
He was just finishing 'The Master Algorithm' when his shabby grandfather clock chimed six o'clock. Time to call Michael. Crammed into his tiny study, Gerry pulled his computer's webcam close and waved at Michael's smiling face.
"Hi Dad! How're you today?"
"I'm alright, son. How're things in sunny Australia?"
"Hot as always. How's things in Scotland?"
"I'd 'ave more chance gettin' a tan from this computer screen than I do goin' out there."
Michael chuckled. He's got that hearty Ferguson laugh, Gerry thought.
"How's the project coming along?" Michael asked. "Am I going to see it one of these days?"
"Of course," grinned Gerry, "I designed it for you."
Gerry's secret project had been in the works for two years now, ever since they found the growth. He had decided it was better not to tell Michael. He would only worry.
The two men chatted for hours. They discussed Michael's love life (or lack thereof), memories of days walking in the park, and their shared passion, the unending woes of Rangers Football Club. It wasn't until Michael said his goodbyes that Gerry noticed he'd been sitting in the dark for the best part of three hours, his mesh curtains casting a dim orange glow across the room from the street light outside. Time to get back to work.
*
Every night, Gerry sat at his computer, crawling forums, nourishing his project, feeding his knowledge and debating with other programmers. Even at age 82, Gerry knew more than most about algorithms. Never wanting to feel old, and with all the kids so adept at this digital stuff, Gerry figured he should give the Internet a try too. Besides, it kept his brain active and restored some of the sociability he'd lost in the previous decades as old friends passed away and the physical scope of his world contracted.
This night, like every night, Gerry worked away into the wee hours. His back would ache come morning, but this was the only time he truly felt alive these days. From his snug red brick home in Scotland, Gerry could share thoughts and information with strangers from all over the world. It truly was a miracle of modern science!
*
The next day, Gerry woke to the warm amber sun seeping in between a crack in the curtains. Like every morning, his thoughts took a little time to come into focus. Instinctively his hand went to the other side of the bed. Nobody there. Of course; she was gone. Rita, the sweetest woman he'd ever known. Four years this spring, God rest her soul.
Puttering around the cramped kitchen, Gerry heard a knock at the door. Who could that be? He could see two women standing in the hallway, their bodies contorted in the fisheye glass of the peephole. One looked familiar, but Gerry couldn't be sure. He fiddled with the locks and pulled the door open.
"Hi Gerry. How are you today?"
"Fine, thanks," he muttered, still searching his mind for where he'd seen her face before.
Noting the confusion in his eyes, the woman proffered a hand. "Alice, Alice Corgan. I pop round every now and again to check on you."
It clicked. "Ah aye! Come in, come in. Lemme get ya a cuppa." Gerry turned and shuffled into the flat.
As Gerry set about his tiny kitchen, Alice called from the living room, "This is Mandy. She's a care worker too. She's going to pay you occasional visits if that's alright with you."
Gerry poked his head around the doorway. "I'll always welcome a beautiful young lady in ma home. Though, I've tae warn you I'm a married man, so no funny business." He winked and ducked back into the kitchen.
Alice turned to Mandy with a grin. "He's a good man, our Gerry. You'll get along just fine." She lowered her voice. "As I said, with the Alzheimer's, he has to be reminded to take his medication, but he's still mostly self-sufficient. We installed a medi-bot to remind him every day and dispense the pills. If he doesn't respond, we'll get a message to send someone over."
Mandy nodded and scribbled notes in a pad.
"When I'm gone, Michael will have somethin' to remember me by."
"Also, and this is something we've been working on for a few months now, Gerry is convinced he has something…" her voice trailed off. "He thinks he has cancer. Now, while the Alzheimer's may affect his day-to-day life, it's not at a stage where he needs to be taken into care. The last time we went for a checkup, the doctor couldn't find any sign of cancer. I think it stems from--"
Gerry shouted from the other room: "Does the young lady take sugar?"
"No, I'm fine thanks," Mandy called back.
"Of course you don't," smiled Gerry. "Young lady like yersel' is sweet enough."
*
The following week, Mandy arrived early at Gerry's. He looked unsure at first, but he invited her in.
Sitting on the sofa nurturing a cup of tea, Alice tried to keep things light. "So what do you do in your spare time, Gerry?"
"I've got nothing but spare time these days, even if it's running a little low."
"Do you have any hobbies?"
"Yes actually." Gerry smiled. "I'm makin' a computer program."
Alice was taken aback. She knew very little about computers herself. "What's the program for?" she asked.
"Well, despite ma appearance, I'm no spring chicken. I know I don't have much time left. Ma son, he lives down in Australia now, he worked on a computer program that uses AI - that's artificial intelligence - to imitate a person."
Alice still looked confused, so Gerry pressed on.
"Well, I know I've not long left, so I've been usin' this open source code to make ma own for when I'm gone. I've already written all the code. Now I just have to add the things that make it seem like me. I can upload audio, text, even videos of masel'. That way, when I'm gone, Michael will have somethin' to remember me by."
Mandy sat there, stunned. She had no idea anybody could do this, much less an octogenarian from his small, ramshackle flat in Glasgow.
"That's amazing Gerry. I'd love to see the real thing when you're done."
"O' course. I mean, it'll take time. There's so much to add, but I'll be happy to give a demonstration."
Mandy sat there and cradled her mug. Imagine, she thought, being able to preserve yourself, or at least some basic caricature of yourself, forever.
*
As the weeks went on, Gerry slowly added new shades to his coded double. Mandy would leaf through the dusty photo albums on Gerry's bookcase, pointing to photos and asking for the story behind each one. Gerry couldn't always remember but, when he could, the accompanying stories were often hilarious, incredible, and usually a little of both. As he vividly recounted tales of bombing missions over Burma, trips to the beach with a young Michael and, in one particularly interesting story, giving the finger to Margaret Thatcher, Mandy would diligently record them through a Dictaphone to be uploaded to the program.
Gerry loved the company, particularly when he could regale the young woman with tales of his son Michael. One day, as they sat on the sofa flicking through a box of trinkets from his days as a travelling salesman, Mandy asked why he didn't have a smartphone.
He shrugged. "If I'm out 'n about then I want to see the world, not some 2D version of it. Besides, there's nothin' on there for me."
Alice explained that you could get Skype on a smartphone: "You'd be able to talk with Michael and feed the geese at the park at the same time," she offered.
Gerry seemed interested but didn't mention it again.
"Only thing I'm worried about with ma computer," he remarked, "is if there's another power cut and I can't call Michael. There's been a few this year from the snow 'n I hate not bein' able to reach him."
"Well, if you ever want to use the Skype app on my phone to call him you're welcome," said Mandy. "After all, you just need to add him to my contacts."
Gerry was flattered. "That's a relief, knowing I won't miss out on calling Michael if the computer goes bust."
*
Then, in early spring, just as the first green buds burst forth from the bare branches, Gerry asked Mandy to come by. "Bring that Alice girl if ya can - I know she's excited to see this too."
The next day, Mandy and Alice dutifully filed into the cramped study and sat down on rickety wooden chairs brought from the living room for this special occasion.
An image of Gerry, somewhat younger than the man himself, flashed up on the screen.
With a dramatic throat clearing, Gerry opened the program on his computer. An image of Gerry, somewhat younger than the man himself, flashed up on the screen.
The room was silent.
"Hiya Michael!" AI Gerry blurted. The real Gerry looked flustered and clicked around the screen. "I forgot to put the facial recognition on. Michael's just the go-to name when it doesn't recognize a face." His voice lilted with anxious excitement. "This is Alice," Gerry said proudly to the camera, pointing at Alice, "and this is Mandy."
AI Gerry didn't take his eyes from real Gerry, but grinned. "Hello, Alice. Hiya Mandy." The voice was definitely his, even if the flow of speech was slightly disjointed.
"Hi," Alice and Mandy stuttered.
Gerry beamed at both of them. His eyes flitted between the girls and the screen, perhaps nervous that his digital counterpart wasn't as polished as they'd been expecting.
"You can ask him almost anything. He's not as advanced as the ones they're making in the big studios, but I think Michael will like him."
Alice and Mandy gathered closer to the monitor. A mute Gerry grinned back from the screen. Sitting in his wooden chair, the real Gerry turned to his AI twin and began chattering away: "So, what do you think o' the place? Not bad eh?"
"Oh aye, like what you've done wi' it," said AI Gerry.
"Gerry," Alice cut in. "What did you say about Michael there?"
"Ah, I made this for him. After all, it's the kind o' thing his studio was doin'. I had to clear some space to upload it 'n show you guys, so I had to remove Skype for now, but Michael won't mind. Anyway, Mandy's gonna let me Skype him from her phone."
Mandy pulled her phone out and smiled. "Aye, he'll be able to chat with two Gerry's."
Alice grabbed Mandy by the arm: "What did you tell him?" she whispered, her eyes wide.
"I told him he can use my phone if he wants to Skype Michael. Is that okay?"
Alice turned to Gerry, who was chattering away with his computerized clone. "Gerry, we'll just be one second, I need to discuss something with Mandy."
"Righto," he nodded.
Outside the room, Alice paced up and down the narrow hallway.
Mandy could see how flustered she was. "What's wrong? Don't you like the chatbot? I think it's kinda c-"
"Michael's dead," Alice spluttered.
"What do you mean? He talks to him all the time."
Alice sighed. "He doesn't talk to Michael. See, a few years back, Michael found out he had cancer. He worked for this company that did AI chatbot stuff. When he knew he was dying he--" she groped in the air for the words-- "he built this chatbot thing for Gerry, some kind of super-advanced AI. Gerry had just been diagnosed with Alzheimer's and I guess Michael was worried Gerry would forget him. He designed the chatbot to say he was in Australia to explain why he couldn't visit."
"That's awful," Mandy granted, "but I don't get what the problem is. I mean, surely he can show the AI Michael his own chatbot?"
"No, because you can't get the AI Michael on Skype. Michael just designed the program to look like Skype."
"But then--" Mandy went silent.
"Michael uploaded the entire AI to Gerry's computer before his death. Gerry didn't delete Skype. He deleted the AI Michael."
"So… that's it? He-he's gone?" Mandy's voice cracked. "He can't just be gone, surely he can't?"
The women stood staring at each other. They looked to the door of the study. They could still hear Gerry, gabbing away with his cybercopy.
"I can't go back in there," muttered Mandy. Her voice wavered as she tried to stem the misery rising in her throat.
Alice shook her head and paced the floor. She stopped and stared at Mandy with grim resignation. "We don't have a choice."
When they returned, Gerry was still happily chatting away.
"Hiya girls. Ya wanna ask my handsome twin any other questions? If not, we could get Michael on the phone?"
Neither woman spoke. Gerry clapped his hands and turned gaily to the monitor again: "I cannae wait for ya t'meet him, Gerry. He's gonna be impressed wi' you."
Alice clasped her hands to her mouth. Tears welled in the women's eyes as they watched the old man converse with his digital copy. The heat of the room seemed to swell, becoming insufferable. Mandy couldn't take it anymore. She jumped up, bolted to the door and collapsed against a wall in the hallway. Alice perched on the edge of her seat in a dumb daze, praying for the floor to open and swallow the contents of the room whole.
Oblivious, Gerry and his echo babbled away, the blue glow of the screen illuminating his euphoric face. "Just wait until y'meet him Gerry, just wait."
Researchers Behaving Badly: Known Frauds Are "the Tip of the Iceberg"
Disgraced stem cell researcher and celebrity surgeon Paolo Macchiarini.
Last week, the whistleblowers in the Paolo Macchiarini affair at Sweden's Karolinska Institutet went on the record here to detail the retaliation they suffered for trying to expose a star surgeon's appalling research misconduct.
Scientific fraud of the type committed by Macchiarini is rare, but studies suggest that it's on the rise.
The whistleblowers had discovered that in six published papers, Macchiarini falsified data, lied about the condition of patients and circumvented ethical approvals. As a result, multiple patients suffered and died. But Karolinska turned a blind eye for years.
Scientific fraud of the type committed by Macchiarini is rare, but studies suggest that it's on the rise. Just this week, for example, Retraction Watch and STAT together broke the news that a Harvard Medical School cardiologist and stem cell researcher, Piero Anversa, falsified data in a whopping 31 papers, which now have to be retracted. Anversa had claimed that he could regenerate heart muscle by injecting bone marrow cells into damaged hearts, a result that no one has been able to duplicate.
A 2009 study published in the Public Library of Science (PLOS) found that about two percent of scientists admitted to committing fabrication, falsification or plagiarism in their work. That's a small number, but up to one third of scientists admit to committing "questionable research practices" that fall into a gray area between rigorous accuracy and outright fraud.
These dubious practices may include misrepresentations, research bias, and inaccurate interpretations of data. One common questionable research practice entails formulating a hypothesis after the research is done in order to claim a successful premise. Another highly questionable practice that can shape research is ghost-authoring by representatives of the pharmaceutical industry and other for-profit fields. Still another is gifting co-authorship to unqualified but powerful individuals who can advance one's career. Such practices can unfairly bolster a scientist's reputation and increase the likelihood of getting the work published.
The above percentages represent what scientists admit to doing themselves; when they evaluate the practices of their colleagues, the numbers jump dramatically. In a 2012 study published in the Journal of Research in Medical Sciences, researchers estimated that 14 percent of other scientists commit serious misconduct, while up to 72 percent engage in questionable practices. While these are only estimates, the problem is clearly not one of just a few bad apples.
In the PLOS study, Daniele Fanelli says that increasing evidence suggests the known frauds are "just the 'tip of the iceberg,' and that many cases are never discovered" because fraud is extremely hard to detect.
Essentially everyone wants to be associated with big breakthroughs, and they may overlook scientifically shaky foundations when a major advance is claimed.
In addition, it's likely that most cases of scientific misconduct go unreported because of the high price of whistleblowing. Those in the Macchiarini case showed extraordinary persistence in their multi-year campaign to stop his deadly trachea implants, while suffering serious damage to their careers. Such heroic efforts to unmask fraud are probably rare.
To make matters worse, there are numerous players in the scientific world who may be complicit in either committing misconduct or covering it up. These include not only primary researchers but co-authors, institutional executives, journal editors, and industry leaders. Essentially everyone wants to be associated with big breakthroughs, and they may overlook scientifically shaky foundations when a major advance is claimed.
Another part of the problem is that it's rare for students in science and medicine to receive an education in ethics. And studies have shown that older, more experienced and possibly jaded researchers are more likely to fudge results than their younger, more idealistic colleagues.
So, given the steep price that individuals and institutions pay for scientific misconduct, what compels them to go down that road in the first place? According to the JRMS study, individuals face intense pressures to publish and to attract grant money in order to secure teaching positions at universities. Once they have acquired positions, the pressure is on to keep the grants and publishing credits coming in order to obtain tenure, be appointed to positions on boards, and recruit flocks of graduate students to assist in research. And not to be underestimated is the human ego.
Paolo Macchiarini is an especially vivid example of a scientist seeking not only fortune, but fame. He liberally (and falsely) claimed powerful politicians and celebrities, even the Pope, as patients or admirers. He may be an extreme example, but we live in an age of celebrity scientists who bring huge amounts of grant money and high prestige to the institutions that employ them.
The media plays a significant role in both glorifying stars and unmasking frauds. In the Macchiarini scandal, the media first lifted him up, as in NBC's laudatory documentary, "A Leap of Faith," which painted him as a kind of miracle-worker, and then brought him down, as in the January 2016 documentary, "The Experiments," which chronicled the agonizing death of one of his patients.
Institutions can also play a crucial role in scientific fraud by putting more emphasis on the number and frequency of papers published than on their quality. The whole course of a scientist's career is profoundly affected by something called the h-index. This is a number based on both the frequency of papers published and how many times the papers are cited by other researchers. Raising one's ranking on the h-index becomes an overriding goal, sometimes eclipsing the kind of patient, time-consuming research that leads to true breakthroughs based on reliable results.
Universities also create a high-pressured environment that encourages scientists to cut corners. They, too, place a heavy emphasis on attracting large monetary grants and accruing fame and prestige. This can lead them, just as it led Karolinska, to protect a star scientist's sloppy or questionable research. According to Dr. Andrew Rosenberg, who is director of the Center for Science and Democracy at the U.S.-based Union of Concerned Scientists, "Karolinska defended its investment in an individual as opposed to the long-term health of the institution. People were dying, and they should have outsourced the investigation from the very beginning."
Having institutions investigate their own practices is a conflict of interest from the get-go, says Rosenberg.
Scientists, universities, and research institutions are also not immune to fads. "Hot" subjects attract grant money and confer prestige, incentivizing scientists to shift their research priorities in a direction that garners more grants. This can mean neglecting the scientist's true area of expertise and interests in favor of a subject that's more likely to attract grant money. In Macchiarini's case, he was allegedly at the forefront of the currently sexy field of regenerative medicine -- a field in which Karolinska was making a huge investment.
The relative scarcity of resources intensifies the already significant pressure on scientists. They may want to publish results rapidly, since they face many competitors for limited grant money, academic positions, students, and influence. The scarcity means that a great many researchers will fail while only a few succeed. Once again, the temptation may be to rush research and to show it in the most positive light possible, even if it means fudging or exaggerating results.
Though the pressures facing scientists are very real, the problem of misconduct is not inevitable.
Intense competition can have a perverse effect on researchers, according to a 2007 study in the journal Science of Engineering and Ethics. Not only does it place undue pressure on scientists to succeed, it frequently leads to the withholding of information from colleagues, which undermines a system in which new discoveries build on the previous work of others. Researchers may feel compelled to withhold their results because of the pressure to be the first to publish. The study's authors propose that more investment in basic research from governments could alleviate some of these competitive pressures.
Scientific journals, although they play a part in publishing flawed science, can't be expected to investigate cases of suspected fraud, says the German science blogger Leonid Schneider. Schneider's writings helped to expose the Macchiarini affair.
"They just basically wait for someone to retract problematic papers," he says.
He also notes that, while American scientists can go to the Office of Research Integrity to report misconduct, whistleblowers in Europe have no external authority to whom they can appeal to investigate cases of fraud.
"They have to go to their employer, who has a vested interest in covering up cases of misconduct," he says.
Science is increasingly international. Major studies can include collaborators from several different countries, and he suggests there should be an international body accessible to all researchers that will investigate suspected fraud.
Ultimately, says Rosenberg, the scientific system must incorporate trust. "You trust co-authors when you write a paper, and peer reviewers at journals trust that scientists at research institutions like Karolinska are acting with integrity."
Without trust, the whole system falls apart. It's the trust of the public, an elusive asset once it has been betrayed, that science depends upon for its very existence. Scientific research is overwhelmingly financed by tax dollars, and the need for the goodwill of the public is more than an abstraction.
The Macchiarini affair raises a profound question of trust and responsibility: Should multiple co-authors be held responsible for a lead author's misconduct?
Karolinska apparently believes so. When the institution at last owned up to the scandal, it vindictively found Karl Henrik-Grinnemo, one of the whistleblowers, guilty of scientific misconduct as well. It also designated two other whistleblowers as "blameworthy" for their roles as co-authors of the papers on which Macchiarini was the lead author.
As a result, the whistleblowers' reputations and employment prospects have become collateral damage. Accusations of research misconduct can be a career killer. Research grants dry up, employment opportunities evaporate, publishing becomes next to impossible, and collaborators vanish into thin air.
Grinnemo contends that co-authors should only be responsible for their discrete contributions, not for the data supplied by others.
"Different aspects of a paper are highly specialized," he says, "and that's why you have multiple authors. You cannot go through every single bit of data because you don't understand all the parts of the article."
This is especially true in multidisciplinary, translational research, where there are sometimes 20 or more authors. "You have to trust co-authors, and if you find something wrong you have to notify all co-authors. But you couldn't go through everything or it would take years to publish an article," says Grinnemo.
Though the pressures facing scientists are very real, the problem of misconduct is not inevitable. Along with increased support from governments and industry, a change in academic culture that emphasizes quality over quantity of published studies could help encourage meritorious research.
But beyond that, trust will always play a role when numerous specialists unite to achieve a common goal: the accumulation of knowledge that will promote human health, wealth, and well-being.
[Correction: An earlier version of this story mistakenly credited The New York Times with breaking the news of the Anversa retractions, rather than Retraction Watch and STAT, which jointly published the exclusive on October 14th. The piece in the Times ran on October 15th. We regret the error.]