Upselling mammograms with AI
I got a call from a relative last week. She’d turned 50 and was scheduled for her first mammogram. She was offered an enhanced “AI mammogram” for an additional $40 out-of-pocket. She asked if she should opt for it.
Why wouldn’t she take advantage of the latest technology to obtain a more accurate interpretation of her scan?
“Say yes to enhanced breast cancer detection when scheduling your screening mammogram” touts the website of one of New York City’s biggest radiology chains. “We are transforming breast cancer detection with a revolutionary offering called Enhanced Breast Cancer Detection (EBCD). EBCD is the only solution of its kind—a package of breast care tools that work in concert to optimize your annual breast cancer screening exam. If you want greater confidence in your results, schedule your screening mammogram appointment today and say YES to EBCD!”
They explain: “EBCD uses Saige-Dx, artificial intelligence technology powered by DeepHealth, to improve the overall cancer detection rate as evidenced in a recent clinical study.”
“It’s like having the world’s most experienced specialists consult on your mammogram,” they claim.
It sounded good until I read a well-reasoned op-ed in the latest edition of the New England Journal of Medicine.
The editorialists note that “as of March 2025, 1.5 million patients had used EBCD. Yet there is no evidence that the addition of EBCD to standard mammography is associated with clinically meaningful benefits.” [my emphasis]
Why not? Detecting more cancers with increased precision may merely create more false positives, necessitating unnecessary biopsies, undue anxiety, and even overly aggressive therapy for slow-growing cancers in older women with limited life expectancy. No clinical trials have assessed the advantage of AI-enhanced mammograms for averting aggressive cancers or reducing death.
EBCD is the latest iteration of a disturbing trend. In 1998, an upgrade to ordinary mammograms (CAD) was rushed into implementation, initially with insurance coverage, until its efficacy was challenged. It, too, did not yield any clinically meaningful results.
3D digital mammography (DBT) came into vogue in the 2000s, and now more than 90% of mammography facilities have it. But again, “clinical efficacy of DBT hasn’t been established” according to the NEJM. Moreover:
“It’s unknown whether observed increases in cancer detection lead to reduced breast cancer mortality or represent primarily overdiagnosis.”
What’s more, DBT doubles the amount of radiation per screening.
Some studies have suggested that the cumulative risk of radiation-induced breast cancer from a lifetime of annual screenings, while small, is substantial.
Large-scale ongoing trials aim to answer the question “Does AI reduce breast cancer deaths?” but until they’re published, the NEJM concludes:
“AI reading aids could represent a clinically relevant innovation—or they could cause harm and be an inefficient use of health care resources.”
So I advised my relative to forego the AI mammogram upsell.
A Smartphone “Fast” Can Make You Smarter
There’s been a lot of buzz these days about how the advent of smart phones has ushered in a host of mental health woes, especially among kids: Decreased attention spans, as well as soaring rates of ADHD, depression, anxiety and even gender dysphoria. Physical health has declined, too, with spiraling childhood obesity and type 2 diabetes.
These have been convincingly linked to the introduction of internet-capable mobile devices in the 2000s, with which people can access addictive social media. Screen time has infiltrated our daily routines as we compulsively scroll on our devices
Jonathan Haidt, in his #1 bestselling book, The Anxious Generation, has thoroughly documented the harms of smartphone ubiquity and is spearheading an initiativecalledletgrow.org:
“If parents don’t replace screen time with real world experiences involving friends and independent activity, banning phones will feel like deprivation, not opening up a world of experiences.”
But in science, the mere concordance of two events is not proof that the predicate was the true antecedent: “Correlation is not causation”.
There was still no conclusive evidence that smartphones were undermining our focus and our moods. That is, until now.
A first-of-its-kind study is forward-looking (prospective), not backward-looking (observational):
“Smartphones enable people to access the online world from anywhere at any time. Despite the benefits of this technology, there is growing concern that smartphone use could adversely impact cognitive functioning and mental health. Correlational and anecdotal evidence suggests that these concerns may be well-founded, but causal evidence remains scarce. We conducted a month-long randomized controlled trial to investigate how removing constant access to the internet through smartphones might impact psychological functioning.”
After a mere one-month smartphone “fast” (participants were allowed to keep their phones, but only for calls and texts):
“The intervention improved mental health, subjective well-being, and objectively measured ability to sustain attention; 91% of participants improved on at least one of these outcomes . . . when people did not have access to mobile internet, they spent more time socializing in person, exercising, and being in nature.”
The study authors conclude:
“These results provide causal evidence that blocking mobile internet can improve important psychological outcomes, and suggest that maintaining the status quo of constant connection to the internet may be detrimental to time use, cognitive functioning, and well-being.”
I think this study is a game-changer, akin to the earliest research conclusively demonstrating cigarettes cause cancer and heart disease.
It’s sure to prompt more concrete action to curb screen time, especially for kids.
And While We’re Bashing Tech . . .
Parkinson’s disease is hard to diagnose in its early stages. Many patients have come to me worried because their hands shake, or their heads bob, wondering if it’s a harbinger of a descent into a devastating neurodegenerative disease.
Indeed, it’s hard even for experienced neurologists to make the distinction between early manifestations of Parkinson’s and benign tremor. The latter can be a nuisance and result in clattering tea saucers but is not a portent of severe debilitation.
There isn’t a single blood test or brain scan can that can clinch the diagnosis of Parkinson’s. That’s why a recent storymade news:
“AI sniffs earwax and detects Parkinson’s with 94% accuracy”
Using a sample of earwax, a chemical analyzer was able to read patterns of volatile organic compounds and compare them to a library of AI-analyzed records to find subjects with Parkinson’s.
But why go high-tech? I seem to recall a story about a woman who could smell Parkinson’s.
And dogs, with their hyper-acute olfactory apparatus, can be trained to detect Parkinson’s—with 98.3% accuracy—years before symptoms become visible and persistent.
Excuse me, but I’d rather have a cute and cuddly animal at my beside than a cold mechanical device. Let’s hear it for DOG scans!
