麻豆传媒

(Image: Michael Kirkham)

I WAKE up feeling lousy. The app on my phone tells me I had a fitful night鈥檚 sleep, which might have something to do with it. But I feel worse than just tired, so I go to the bathroom and pee on a sensor strip. Most metabolites are fine but there鈥檚 an excess of nitrites, which could indicate a urinary tract infection.

Back in my bedroom I run more tests. My blood results tell me my vitamin D levels could be higher 鈥� but I live in dreary London, so I knew that already. My heart rate and blood oxygen are fine, though my inflammation levels are higher than normal. Then I swab my nose and the machine by my bedside takes just a few minutes to tell me what鈥檚 really wrong: I鈥檝e got the flu. A quick look at the flu map on my phone tells me that 8 per cent of the neighbourhood has succumbed. I鈥檓 just another red dot.

The way illness is diagnosed is changing, and we are all invited to take part. Each of the technologies described above is on the market already, or will be soon 鈥� and there are a lot more where those came from.

Some will feel familiar to people who already monitor their health with smartphone apps or wearable devices. But there鈥檚 a difference: instead of merely tracking lifestyle indicators such as sleep quality, diet and physical activity, they will also deliver medical diagnoses and advice. Welcome to the age of DIY diagnostics.

Apps and wearables are already moving in this direction. You can buy an elecrocardiogram to attach to your smartphone, for example, and top of the range fitness trackers can now log your pulse and respiration rate; the next generation will also track blood pressure, glucose levels, hydration and blood oxygen.

Next will come a range of gadgets that bring professional-level diagnostic equipment into the home. In December, San Diego company the first generation of its bedside-table device, a little white box that can diagnose flu and measure four vital signs: vitamin D levels, inflammation, female fertility and testosterone (see case study: cue).

(Image: Cue)

Meanwhile, the 10 finalists of the $10 million Tricorder X Prize started testing their Star Trek-style 鈥減ocket doctors鈥� on people last month. The prize will be awarded to the first hand-held device that enables somebody without medical training to diagnose 15 medical conditions and monitor five vital signs (see 鈥�Tricorder tests鈥�). The winner will be announced in January and it won鈥檛 be long before the devices are on sale to the public.

Those behind this emerging industry say the goal is a healthier, more engaged population who visit their doctor less often 鈥� and when they do, arrive with a diagnosis and data to back it up. 鈥淭his will bring about a profound change in the way healthcare is delivered,鈥� says Ali Parsa, chief executive of digital healthcare company Babylon.

But others fear we are asking for trouble. How reliable are the tests? What if, rather than keeping people away from the doctor鈥檚 office, they cause a stampede towards it? And who owns the data that this diagnostic boom is going to produce?

The first home diagnostic was a pregnancy test launched in the US in 1978. Today you can buy dozens of home tests online or from pharmacies for anything from infectious diseases such as TB to the onset of menopause.

However, getting reliable information about how good the tests are isn鈥檛 easy. A few have been approved by health authorities, but many more have not. In 2011, UK magazine Which? asked two doctors and a panel of 146 users to . They concluded that people would be better off saving their money and going straight to their doctor, who would run their own tests anyway.

The newcomers want to change that. 鈥淐urrent consumer diagnostics aren鈥檛 very satisfying,鈥� says Eugene Chan, head scientist at the DNA Medicine Institute (DMI), one of the companies shortlisted for the Tricorder X Prize (see case study: rHEALTH). 鈥淭hey need to be confirmed by a more sophisticated method鈥� 鈥� which is exactly what DMI and its competitors are aiming to provide.

Then there鈥檚 the fact that existing tests generally diagnose just one specific condition. That鈥檚 fine if you already have an inkling what鈥檚 wrong, but what if you don鈥檛? 鈥淭he tricorder devices are designed for people who may not know what is wrong with them,鈥� says Grant Campany, senior director of the Tricorder prize. In that sense, the devices work more like a doctor: they ask questions about symptoms and then run tests to reach a plausible diagnosis.

Exceptional opportunity

All this is being made possible by advances in artificial intelligence, wireless sensing, diagnostic imaging and lab-on-a-chip technology, as well as the ability to shrink it all down into a compact device. Couple that with the increasing sophistication of smartphones and cloud computing, which allow medical data to be stored and shared, and you have 鈥渁n exceptional technical ecosystem for healthcare鈥�, says Campany.

That鈥檚 the marketing pitch anyway. But there鈥檚 many a slip between pee cup and stick.

An obvious concern is accuracy. If home diagnostics produce ambiguous, hard-to-interpret results or high rates of false positives, the upshot is likely to be even greater demand for medical appointments. Similarly, a high rate of false negatives might misleadingly reassure people and cause them to delay seeing a doctor when they really need to. We have seen the problems that both types of error can cause with population-wide screening programmes, which have largely been withdrawn or scaled back because, overall, they do more harm than good.

A related issue is how people respond to the results. Being diagnosed with a serious disease can be traumatic; without professional help to put the result in context, it can be even worse.

Ideally, DIY diagnostics should be tested in clinical trials, says Annette Pl眉ddemann, director of the Diagnostic Horizon Scan Programme at the University of Oxford 鈥� both to assess their accuracy and to ensure that they help people make sensible decisions. Yet few such trials have been done for the tests on the market, she says. Exceptions include a recently approved in the UK and US, and and .

The trouble is, manufacturers don鈥檛 have to carry out clinical trials to gain approval to sell their products. 鈥淲e need a better regulatory process,鈥� says Chris Price, an expert in diagnostics at the University of Oxford. Much of the current UK regulation assesses how technically good a test is, but not whether it helps improve the user鈥檚 health, he says.

Accuracy is already an issue for smartphone apps. For example, an designed to calculate how much insulin people with diabetes should take based on their blood sugar and the carbohydrate content of their next meal discovered that two-thirds carried a risk of inappropriate dosing.

The makers of next-generation diagnostics are confident that they can do better. The US regulatory process for such devices is still a work in progress, but Campany says he expects the Food and Drug Administration to demand clinical trials to validate the safety and accuracy of tricorders before they go on sale.

Chan agrees. 鈥淲e鈥檙e aiming to have the same gold standard as laboratory diagnostic tests,鈥� he says. Ayub Khattak, CEO of Cue, is similarly confident: 鈥淲e do a lot of testing relative to standard lab tests and we鈥檙e very good,鈥� he says.

DIY diagnostics could make tests like these a thing of the past (Image: Renee Keith/Getty)

Assuming that diagnoses are good enough, the next task will be to ensure they lead to an appropriate medical follow-up. 鈥淚t鈥檚 naive to think that the technology by itself is enough 鈥� you need the support package,鈥� says Jeremy Wyatt, head of e-health research at the University of Leeds, UK. That might just mean a prescription or referral to a specialist. But it could be more involved, such as counselling or 24-hour support.

For that to happen, DIY diagnostics will have to be accepted by doctors, though this doesn鈥檛 look like a major hurdle. Doctors in the UK already encourage people to monitor their own blood pressure and other indicators, and the National Institute for Health and Care Excellence has endorsed self-monitoring for people with diabetes and those on blood thinning drugs. Widening the net to self-diagnosis isn鈥檛 too much of a step into the unknown. 鈥淚f the tests are of suitable analytical quality and they have been shown to be clinically useful, I don鈥檛 see why doctors wouldn鈥檛 support them,鈥� says Price.

In a survey of US doctors, researchers at PwC鈥檚 Health Research Institute found that over half said they would be comfortable using vital-signs data from an app to prescribe medication, and 47 per cent said they were happy to prescribe drugs based on a self-administered test.

Good-quality follow up should also help prevent an epidemic of hypochondria similar to the one that struck when people first started Googling their symptoms. In 2008, Eric Horvitz of Microsoft Research in Redmond, Washington, co-authored the first systematic study of this issue. He concluded that medical web searches frequently lead people to think the worst 鈥� a phenomenon called 鈥�.

鈥淲eb searches often lead people to think the worst 鈥� a phenomenon called 肠测产别谤肠丑辞苍诲谤颈补鈥�

鈥淢ost folks are just fine,鈥� says Horvitz. 鈥淏ut with the introduction of these new devices, promising as they are, we will have to balance the value against the cost in dollars and anxiety.鈥�

Consumers are likely to have another anxiety too: ownership and security of their health data. In a recent survey, also carried out by PwC, 76 per cent of people said they were concerned about the security of their medical records and 68 per cent about that of health data stored in smartphone apps.

Again, the companies say they have this covered. For most diagnostics in the pipeline, data will be encrypted, password-protected and stored on the company鈥檚 servers. 鈥淯sers would have their own account for this information, like the way you have your own email address,鈥� says Chan. Details are anonymised and people can decide whether to opt in to share them, for example, on Cue鈥檚 Flu Map. The data is thus as secure as anything else stored in the cloud.

There is clearly still much that we don鈥檛 know about what happens when DIY diagnostics meet the real world. But some are already looking forward to the next step.

鈥淲e will shortly see the advance of artificial intelligence in healthcare,鈥� says Parsa. 鈥淢achines will be learning everything there is to learn about health and medicine.鈥� IBM鈥檚 supercomputer Watson is already helping doctors to diagnose cancer, and will soon be genned up on other conditions. The Tricorder teams are currently using their own diagnostic algorithms, says Campany, but he can envisage Watson being recruited at some point to create an even more powerful device.

Another promising direction is to utilise diagnostic information that is out there but not currently used, says Tim Riedel of the University of Texas at Austin. There鈥檚 already an that diagnoses a type of eye cancer from selfies. What about the colour of your tongue, or changes in the way you type on a keyboard? 鈥淭ricorders are cool, but what would be cooler is if we can harness devices people already have,鈥� says Paul Wicks, head of innovation at patient-support website PatientsLikeMe. 鈥淐all it capturing the data exhaust. It seems a bit less sexy than a new machine that goes 鈥榩ing鈥�. But two years ago we were talking about Google Glass and all the applications for that. Look what happened鈥�.

One thing seems clear: whether you are ready for it or not, the way we diagnose diseases is undergoing some serious DIY disruption. And we are just at the tip of the iceberg, says Riedel. 鈥淚 keep telling my students: 鈥業 got my first cellphone when I graduated college. You were handed one in your crib鈥�. They and the generation after them will make diagnosis happen in mind-blowing ways we can鈥檛 even think of right now.鈥�

This article appeared in print under the headline 鈥淒octor you鈥�