Diagnose this

A health-care revolution in the making


Imagine it’s the future. Imagine your future self rolling out of bed in the morning and heading for the bathroom. Your smart toilet is an older model and you are thinking of getting a new one.

Sure, the old toilet can do a basic urinalysis, picking up indicators of incipient diabetes or infection. And it can alert you to blood in your stool, a potential sign of colon cancer, just as quickly as you can flush and squint at the readout. Your special test-strip toilet tissue — “Accurate yet Soft!” — gives you a green thumbs-up on 30 different daily diagnostics. And the toilet reports that your gut microbiome is up to snuff.

But your model doesn’t test for any of the dozen healthful new gut bacteria discovered among African hunter-gatherers.

You took the Kan+™ probiotic capsules; have the microbes colonized your gut yet?

What really has you lusting for a new toilet, though, is the lack of data-share options for your old toilet. Honestly, your doctor and one emergency contact? That’s it? Who’s going to help you make sense of all this information? What about GloMM, the global health record data bank founded in 2021 that stores and shares all your mobile and other health data? What about your two dating sites? A lot of potential partners expect to know how healthy you are. Not to mention SocialWell, which will match the government’s $3,000 rebate if you get a new smart toilet before the end of the year.

Back in the present, we are talking with Sanjiv Sam Gambhir, MD, PhD, who’s working to translate such a scenario — or one a little like it — into reality. Gambhir, chair of Stanford’s Radiology Department and director of the Canary Center at Stanford for Cancer Early Detection, envisions a future where we nearly continuously monitor our health. The resulting data might tell each of us or our health-care team, right away, if something is amiss. Are we developing tiny aggressive tumors? A slight tremor suggestive of the onset of a neurodegenerative disease? Or organ-damaging high blood pressure?

Current diagnostics, says Gambhir, are so intermittent, it’s like trying to watch a movie but seeing it only every 20 to 30 minutes for a few seconds each time until near the end of the movie when you get to watch it for a few minutes. Inevitably, we’ll miss critical parts of the story.

In general, diagnostics have been underappreciated. According to a 2015 National Academy of Medicine report, “The delivery of health care has proceeded for decades with a blind spot: Diagnostic errors — inaccurate or delayed diagnoses — persist throughout all settings of care and continue to harm an unacceptable number of patients.” Gambhir is one of the few who recognize how systemic the problem is, how colossal the challenge, and who want to change things.  

The underpinnings of a greater emphasis on diagnostics will be devices that can monitor health at all times. Radiology lecturer Seung-min Park, PhD, who works in the Gambhir lab, is helping to lay the foundation for Gambhir’s diagnostic vision. If you want to continuously monitor the body, says Park, you can’t do that with anything like surgery, blood draws or X-ray imaging. No one would put up with that.

It is clear, Park says, that the perfect sources of diagnostic information are the molecular contents of sweat, saliva, urine and feces, naturally excreted every day and packed with information. Researchers around the world have realized that these substances can provide clues to our health.

Park is engineering a smart-toilet prototype that can collect urine for testing several times a day. To get the project started, he’s using an off-the-shelf commercial test strip that measures 10 factors such as acidity, which can tell you about your risk of kidney stones, and glucose, an indicator of diabetes.

The Gambhir lab is also working on a smart bra designed to continuously image breast tissue. The bra uses a combination of infrared light and sound to image and detect minuscule breast tumors, so they can be removed long before they metastasize. Like the smart toilet, the smart bra is still under development. For now, the lab’s engineers are scratching their heads over challenges like how to analyze the nonstop flow of data and where to place the battery.

Cardiologists are already making the vision of continuous monitoring a reality. Information from pacemakers and other devices implanted in the heart can be transmitted automatically through ultralow radio frequencies so that patients can be monitored for signs of crisis.

For example, when an infant was born with a deadly heart arrhythmia, her doctors at Lucile Packard Children’s Hospital Stanford implanted a pacemaker and defibrillator in her heart that could report back to her doctors if the defibrillator was activated. At 7 months, the defibrillator began to go off. Although the baby looked fine to her parents, she was in serious trouble. The hospital told the parents to bring the baby in right away, and within a few weeks a heart transplant saved her life.

Gambhir’s vision

Diagnostics have moved far beyond old-​fashioned X-rays for broken bones. We already live in a world where, if we wanted, we could monitor our health around the clock with a variety of ingenious devices that can potentially help foretell illness.

Wearable and implantable devices can deliver rivers of information that can both help health-care systems track the health of individuals and help researchers study the effectiveness of treatments or preventive health programs in whole populations. Some people won’t want to be monitored all the time, Gambhir acknowledges, but he thinks that for many the desire for the benefits will outweigh their concerns about privacy.

Gambhir compares the future of diagnostic medicine to the approach used to keep the engines of commercial jets spinning smoothly and safely. “Most people have taken a flight on a commercial jet,” he says. “You may not know it, but the jet engines on that plane are almost continuously monitored by an engine-health portal that sits at General Electric or Rolls-Royce. Every 30 seconds, each engine on the airplane sends information down to the engine-health portal. Hundreds of sensors built into that jet engine are letting the health portal know if there’s a problem with the engine — even in flight. If there’s a problem, adjustments to the engine can be made, without the pilots even knowing, still in flight.” For more serious problems, a plane can be forced to land. Just as importantly, jet engine engineers have learned when not to intervene and just continue to monitor — to avoid false alarms.

‘The future is all about being able to intercept diseases early and, ideally, prevent them. If we can actually do something about a disease such as an aggressive cancer, then it is worth monitoring for it.’

“There is no real equivalent in health care,” says Gambhir. “There isn’t a continuous monitoring of your health. The future is all about being able to intercept diseases early and, ideally, prevent them. If we can actually do something about a disease such as an aggressive cancer, then it is worth monitoring for it.”

Yet when research dollars are doled out, diagnostic tools are often treated as an afterthought, Gambhir says. People don’t think of diagnostics as saving lives, but treatment depends heavily on accurate diagnosis — and biomedical research even more so. Expenditures on the field of diagnostics research are not tracked separately, but he estimates that no more than 7 percent of total biomedical research dollars go to diagnostics, with the rest going to discovering ever more treatments.

Gambhir would love to see that ratio reversed, he says, so that the “anticipating and preventing disease” part of Stanford’s precision health approach takes priority over endless new treatments.

But he concedes he’d be happy with a 50:50 funding split between diagnostics and therapeutics and anticipates such a transition in the coming years. It makes much more sense, he argues, to put resources into preventing disease or at least diagnosing disease early — when, in many cases, it’s easier to treat — than doing nothing until people are quite ill.

But the way biomedical research is funded and the way medicine is practiced are still structured around treatment, not diagnosis. So a diagnostics-first approach would mean major changes.

The structure of medicine

Kathryn McDonald, the executive director of Stanford’s Center for Health Policy and the Center for Primary Care and Outcomes Research, concurs with Gambhir that diagnostics are severely under­studied, given how important they are. “Our health-care system is organized around what happens once you already know what’s wrong, as opposed to figuring out what’s wrong,” McDonald says.

In 2015, the National Academy of Medicine reported that at least 5 percent of U.S. outpatients experience a diagnostic error, 6 to 17 percent of adverse events in hospitals result from diagnostic errors, and diagnostic errors contribute to 10 percent of all patient deaths.

Yet, despite the importance of diagnostics, they receive minimal funding, says McDonald, who serves on the National Academy of Medicine’s Committee on Diagnostic Errors in Health Care. “If you look at the dollars associated with diagnostic testing, it just pales in comparison to dollars spent on pharmaceuticals. And there’s a parallel in the research world.”

One reason is that diagnostics is primarily a cognitive activity, McDonald says. It’s your doctor sitting and thinking, reading, thinking some more, calling a colleague and talking until they figure out what’s wrong with you. And there’s almost no support for thinking and talking, she says. Physicians and others are compensated for treating patients and, to a lesser extent, for seeing patients, but not for thinking about them.

We need to look for ways to reward that cognitive work and teamwork, says McDonald.

False positives, false negatives and false reassurance

Although diagnosis may happen through thinking and communicating, diagnostic tests themselves, and how physicians think about them, are susceptible to error. Tests are notorious for generating false positives and false negatives, and the more rare the condition, the easier it is to be misled by such false information.

In the case of a test for blood in the urine, a false positive would indicate there was blood when there wasn’t actually blood there. Likewise, a false negative would essentially be a miss; the test result would say there is no blood when in fact there is.

False positives can generate a lot of anxiety for patients and waste health-care dollars for everyone. But besides the problem of false positives and negatives, McDonald also points out that continuous monitoring could be prone to false reassurance. If you are using a smart toilet or smart bra, she says, you might decide you don’t need a regular lab test. But the device could stop working, and you might not know it.

The integration piece

Collecting information about ourselves is only a piece of what gets us to better patient care, says Leslie Saxon, MD, professor of clinical medicine at the University of Southern California. Saxon heads the USC Center for Body Computing, a major center for the development of diagnostics.

Diagnostics could be information from wearable devices, says Saxon, a member of the small cadre of researchers interested in what diagnostics can contribute to the future of medicine. “But diagnostics is also what patients are telling me, or what their mother or sister are telling me: ‘He hasn’t gotten out of bed for three days. He’s depressed.’ ” Diagnostics, she says, have to be integrated with everything we know about patients.

For example, information from devices for monitoring heart activity have to be considered in the context of what else we know — whether a patient is taking her prescriptions or how she is using the monitor.

And diagnostics and biomarkers are just a piece of the puzzle, she says. The bigger challenge may be handling that information — processing it, integrating it and sharing it — in a way that helps both patients and researchers.

Not so fast

Peter Schmidt, PhD, senior vice president and chief mission officer at the National Parkinson Foundation, casts a gimlet eye on what he views as overenthusiasm for biomarkers and diagnostics.

It’s not that he’s against diagnosing people who are ill. But for a variety of reasons, not all diseases are good targets for continuous monitoring, he says. Cancer, for example, is an appropriate target for continuous monitoring because it’s typically easy to treat when caught early, difficult or impossible to treat when caught later. But neurodegenerative diseases such as Parkinson’s disease are difficult to treat at all, let alone cure, so knowing you have it before you even feel sick could be a negative.

“A human is not a jet engine and we deal with problems in our own way,” Schmidt says. He questions the wisdom and ethics of diagnosing people with illnesses when they feel fine and when intervention won’t clearly do them any good.

Imagine, he says, that you are 70 years old and have been feeling fine, but a test has just revealed that you have Parkinson’s disease. “You aren’t actually aware of any symptoms, and then you die a year or two later from a heart attack. Having been told you have Parkinson’s disease would have helped you not at all.

“Parkinson’s disease can be completely managed for a year or two after diagnosis,” Schmidt adds. “During that two-year period, Parkinson’s disease is mostly a disease of fear, where people will think, ‘Eventually this disease is going to overcome the effects of the medications, and it is already doing something bad to my brain.’ That’s a scary thing.”

Manifold challenges

Diagnostics encompass far more than just figuring out what is wrong with one patient. If medicine moves toward a more preventive model, that will require better diagnostics. Such a future requires support for research on diagnosis and structural support for timely and accurate diagnosis, says McDonald.

“And,” she says, “the research is not just about training physicians to do a better job. It’s about how the delivery system is supporting them in doing that, how the payment system is supporting them in doing that, how the legal system is supporting them in doing that.”

The number of people looking at how the entire health-care system can support diagnostics is, for now, a “small tribe” of people, says McDonald. “This problem matters. It needs attention, and no one is funding the research to build a knowledge base to help you write your article,” she says with a smile.

As Gambhir emphasizes, the changes, if they come, could take decades, and the challenges are manifold. At one level, he says, the challenge is in understanding both our biology and the output from all these new devices well enough to know what to do with the information. The biology of early disease is not necessarily the same as that of late disease. Another major challenge, says Saxon, is handling and processing and sharing that information in a way that helps patients. And, as McDonald says, “The current health-care system is shaped more for treatment than for diagnosis, more for action than for thinking.”

The smart toilet of the future won’t be a stand-alone device, but part of an integrated network of information about you and billions of other people, in a system — of devices, servers, institutions and individuals — that actively prioritizes diagnosis, communication and prevention. Instead of flushing millions of petabytes of data into the sewers each day, we’ll wrest from it the seeds of a healthier future.

Author headshot

Jennie Dusheck

Jennie Dusheck is a science writer for the medical school's Office of Communication & Public Affairs. Email her at dusheck@stanford.edu.

Email the author