Out in Public
Microsoft's New Smart Bra
The bra market has a whole new category – wearable cancer detection devices. Put out by First Warning Systems of Reno, Nevada, the new bra may not immediately win fashion awards – unlike the Microsoft "Smart Bra" pictured to the left. Yet the magazine “The Week” declared it July 18th’s “Innovation of the Week.”
The new bra – named the Circadian Biometric Recorder – will look at heat and other sensor readings inside the bra over a 12-hour period. Using a series of techniques from “big data,” by tracking the sensors feeds and sending them through proprietary algorithms, it will determine who likely has a growing breast cancer – and needs a breast biopsy. The company reports a "74 percent plus correlation" between cancer disease states and the results of its new bra.
Yes, this is a world of fake news, whose “truthiness” often appears more plausible and appealing than the real thing. Yes, having run the Center for Circadian Medicine for nearly 20 years I might be eager to broadcast the latest circadian “product,” especially since many people can’t pronounce circadian and often don’t know what it means (around a day – representing 24 hour body clock medicine.)
But No, the Circadian Biometric Recorder is real.
Better Than A Mammogram?
Cancer screening produces many, many problems (see below.) Yet the businessmen behind the Circadian Biometric Recorder are setting their sights on a process which needs great improvement – and could provide investors real profits.
Though about one in eight women in America is now getting diagnosed with breast cancer, mammography is a lousy technology. It’s expensive. It scares women – and the people that love them. Practitioners who read the studies fight hard to get an accuracy of 85 percent – even when they do nothing else but read mammograms.
False positives are everywhere. People don’t like getting irradiated. Women hate waiting for the Sword of Damocles to reign over their heads. Physicians detest dealing with the very large group of false positive results.
And the usefulness of mammography screening leaves a lot to be desired. Many epidemiologists view mammography's utility as marginal. Studies, like those described by H. Gordon Welch in “Overdiagnosed” show 20 percent more positives with yearly mammography than with an every five year version – arguing that 20 percent of the tumors disappear on their own. Examining yourself is as effective as yearly mammography in detecting tumors in several studies. And the tumors that mammography misses can still maim and kill – and drive patients, their families and their physicians nuts.
So an inexpensive test where you wear a device at your house, have the results read off the cloud, and don’t worry about radiation has much to commend itself.
If it works.
The Internet Of Things
Home health diagnostics gets Silicon Valley digerati gesticulating feverishly. The money, the safety, the convenience, the “self-empowerment.” They can’t control themselves.
All the more reason to choose caution.
A lot of home health based materials have been rotten. Since much of the public – including experts – know hardly anything about statistics, it’s easy to bamboozle folks.
The “sleep records” people have shown me off their cellphones and fitbits appear terribly inaccurate to a dumb clinician like me. They mistake REM sleep with waking, and often record non-existent events. They results don’t look anything like what comes out of a lab based sleep study. Similarly, numerous genetic testing services have also been criticized for using correlation as explanation. No, you may have less Neanderthal genes than they tell you – and your risk of a heart attack in three years may not be precisely 36 percent.
Scientific information is often thorny, tangled, highly variable from place to place and society to society. Home devices have more than just “glitches” – though these sometimes appear ubiquitous. The results page often does not explain to people how many “outliers” are out there – that real data feeds are often fuzzy, or filled with numbers that look completely out of whack.
Because, statistically, whether you’re using Gaussian or Poisson distributions or others, unusual events are not that unusual. They happen all the time - especially when the sample size is large. Spread over enough people and events, the “unpredictable” event will appear surprisingly predictable – and common.
So your internist may not become too thrilled to wake up at 3 AM and be told your cellphone reports you are in ventricular tachycardia and she needs attend to you in the ER right now. Getting all this stuff to work accurately in prime time – and home time – is more than a challenge. It will prove a huge hassle – for communities, not just individuals.
And that’s before we even consider the security issues. Hacking home health devices has so far proven far too easy. Ask the makers of AICD devices.
The Circadian Biometric Recorder wants to replace the mammogram. The manufacturers argue their device makes a lot of sense in places like Southeast Asia, with high breast cancer rates, low incomes and unsophisticated medical facilities.
But will a heat based device used in a Vietnamese population which never sees air conditioning have the same accuracy as in a carefully selected American population? Will rural Vietnamese mimic the the sensor patterns of obese Californians? Will “Big Data” effortlessly apply its results to new populations without a hitch?
Rightly, First Warning is doing a clinical trial with 175 patients in southern California. But it is comparing them against “historical” mammography data – rather than the real thing.
For double blind trials are necessary to determine if anything works. They are difficult, time consuming and expensive. Plus device companies can often get away with much less careful clinical research than drug companies when getting their wares through European and American regulators.
Nor do clinical trials in the best circumstances consider the other parts of health – the mental, social and spiritual costs – and benefits – of screening and treatment. They don’t track what happens to people psychologically when going through cancer testing. They rarely take attend to the long term physical changes and debilities screening brings – nor the economic and social costs. What happens when that boyfriend you want to make your partner finds out you “might” have breast cancer? Or your mother and father? Nor do trials look at the effects on communities.
Cancer screening presently has a lot to answer for. Device and diagnostic test makers know very well that find a cancer and people immediately want to treat it – right now. Consider the PSA, the prostatic serum antigen test. The supposed originator of the test says he would never have used it for screening. European and American trials showed 30-100 people treated “unnecessarily” for every patient “cured.” What does that mean? That for every prostate cancer “survivor,” there may be 30 or 50 or 100 men who can’t have sex; or dribble when they walk; or are continuously weak because radiation has wasted parts of their GI tract. Even the American Urological Association says men over 70 should not be routinely screened by the PSA. As my patients prove to me, they often are – and are treated no matter how old they may be.
Home health testing will come. It should eventually be cheaper, more efficient, more networked, more accurate. But it has to prove that first.
Good luck to First Warning Systems. They have a worthy project. And a lot to prove before anyone starts wearing a breast cancer detection bra.