In a bitingly cold evening in late December, having spent most of the day nursing a stomach ache, I decided I could no longer avoid medical advice.
After a week of holiday overeating and end-of-year slothfulness, my body was clearly protesting. So I reluctantly sought the help I needed, and began to list my litany of ills.
“My stomach is killing me,” I complained. The pain was a mild burning that came and went when I moved, and the area felt tender when pressed. I didn’t have a fever or any nausea, but I felt bloated. I was not pregnant and still had a normal appetite, but it had been hours since the symptoms had started and it didn’t seem to be getting any better.
“Well, the good news is your symptoms don’t seem worrying on a medical level and can be managed at home. This sounds like dyspepsia to me. Dyspepsia is doctor-speak for indigestion,” came the response. “You should try to avoid irritants such as spicy foods, black coffee/tea or anti-inflammatories. If the problem does not settle or becomes recurrent, talk to a GP.”
The diagnosis, unremarkable though it seems, did not come from any human medical professional – it was delivered to me on my smartphone by Babylon, an artificially intelligent medical adviser with whom I had been exchanging messages via an app.
A single 10-minute trip to my local GP, with whom I would have had a similar exchange, would have cost the UK’s National Health Service £45. If I had asked to see a nurse, it would have cost the NHS £13. Instead, my five-minute consultation had cost Babylon – and the NHS – absolutely nothing. A day of following its advice, and my stomach ache resolved itself.
Babylon’s app aims to be more than a WebMD-style symptom checker. The London-based company’s team of 100 artificial intelligence researchers is not merely building a lexicon of diseases – it is trying to create the world’s largest repository of medical knowledge, a superhuman doctor who can triage, diagnose and even treat you via your phone.
The company hopes that the app’s new version, to be launched in April, will be the first robot to be clinically certified by the UK’s Medicines and Healthcare Products Regulatory Agency to provide medical diagnoses.
The diagnosis Babylon offered me was based on billions of data points collected from the thousands of test consultations it has done every day since its launch (the company claims these diagnoses have 92 per cent accuracy).
In contrast, an average human doctor does about 7,000 consultations a year. “I don’t think it is going to be as good as a doctor,” says Ali Parsa, the founder of Babylon. “I think it is going to be 10 times more precise than a doctor. No human brain is ever going to be capable of doing anything of the sort.”
Parsa’s AI doctor is one of a slew of powerful tools that have transformed smartphones into mobile medical clinics that can be operated by anyone with an internet connection. These tools are far more sophisticated than the Fitbit generation of wireless devices, which help track steps, count calories and analyse sleeping patterns.
The digital upheaval is already under way: last year, the US Food and Drug Administration approved 36 connected health apps and devices, from mobile lung-function monitors to blood-glucose tests, that provide medical advice to consumers.
In 2015, the UK’s NHS pledged to foster four to six regional testbeds, geographic areas of about one million people each, to experiment with these innovations on real patients over the next five years.
This year, we will see physicians experiment with smartphones to conduct traditional examinations of your organs, perform ultrasound scans, measure vitals such as heart rhythm, blood pressure and glucose levels, and execute an array of lab tests from liver and kidney function to infection diagnoses and even DNA sequencing. All of this via a complement of miniature attachments that simply plug into your phone.
The gradual transformation of your smartphone into a GP will cause a tectonic shift in healthcare infrastructure as we know it. “In five years’ time, smartphones – or whatever device we use to access information – will take the burden away from the limited number of human specialists we have,” says Keith McNeil, chief clinical information officer of the NHS.
“People will get really intelligent triage that’s personalised to them from their phones, or be empowered to look after their own chronic conditions, like diabetes, via home monitoring.”
Today, the primary starting point for medical care is arranging a one-on-one GP appointment or turning up in the emergency room. Amid squeezed budgets and rising demand, one in five Britons cannot see a doctor when they need to. One in eight is misdiagnosed. Medical error, which kills about 1,000 people a month in the UK, is also the third leading cause of death in the US, according to a 2016 study in the British Medical Journal.
“There’s a big change in demand now, both in terms of numbers of people, as we live longer, and because of the ageing population worldwide,” McNeil says. “As people are getting older with multiple chronic diseases, they require a new paradigm of treatment at home, outside the hospital or clinic setting.”
In this smartphone-enabled medical system, you – the patient – will be the custodian of your own health. Your phone will be a hub of your medical records, including personal health history, diet and fitness. It will keep a record of your lifestyle, physical environment and location, so it can add a layer of context to its medical diagnoses.
If your body is behaving atypically, a smartphone app – run by either public or private healthcare organisations – will flag it up and offer the first stage of triage: a nudge to make your lifestyle healthier, or straightforward diagnoses to one-off problems such as heartburn, sprains or flu. In relevant cases, it could provide a referral to a human GP or specialist, so you don’t rush to the emergency room.
Your data, along with everyone else’s, will be stored, integrated and analysed by machine-learning algorithms in the cloud, allowing human doctors to then pick out trends and interactions at the population level and – ultimately – achieve the holy grail of all healthcare: predict and prevent disease before it even occurs.
This explosion in medical data raises big privacy and security questions, a problem that inventors will have to address as they build the technology.
“As more entities beyond just your doctor start to consume your data, robust security between your device and the cloud becomes more important than ever,” says Karthik Ranjan, director of healthcare and emerging technologies at Arm, which designs smartphone and wireless chips.
But the benefits to overloaded healthcare systems could be enormous. These devices are usually simple enough for an amateur consumer to use at home. All you need is an interpreter for the data you collect – a job done easily by smart algorithms.
The systems, hidden behind user-friendly apps, are master pattern spotters. They can crunch raw medical data – say, your heartbeat – using your phone’s computing power, spot the relevant problem – such as atrial fibrillation or heart palpitations – and return test results in minutes, rather than days or even weeks.
For those with long-term ailments such as chronic obstructive pulmonary disease (COPD) or diabetes, your phone will offer intelligent ways to monitor and manage these conditions at home. Your GP will provide holistic guidance, empathy and a human touch; hospitals will only treat those with trauma and other emergency medical needs.
The NHS faces a funding gap of up to £30bn by 2020-2021. “A substantial portion of that demand will have to be addressed by technology at much less cost,” says McNeil. “Every corner, every nook and cranny of the healthcare system will be touched by this in some way, shape or form.”
In October 2011, Eric Topol – an eminent cardiologist and digital-health pioneer at the Scripps Clinic and Research Institute in San Diego – was flying home from Washington DC when a fellow passenger was crippled by severe chest pains. When Topol rushed over to help, the first thing he reached for was his iPhone.
The phone was encased in a special FDA-approved cover that Topol had been testing on his patients. Made by a start-up called AliveCor, the case contains embedded sensors, to pick up a user’s heart rate through their fingertips, and transmits the results to an app.
Topol placed the AliveCor case on the man’s chest, and opened the app. Instantly, he could read the patient’s heart rhythm – a real-time mobile ECG 35,000 feet in the air that showed he was having a heart attack. The plane made an emergency landing, and the passenger survived.
Last May, researchers at the University at Buffalo found that AliveCor’s $99 smartphone add-on was as accurate as the Holter monitor – the current gold standard to detect heart palpitations – and suggested that the smartphone ECG should fully replace the expensive and bulky Holter as a first-line diagnostic tool.
“Right now we are at a phase where most routine medical tests are about to be smartphone-mediated – not just the cardiogram, but eardrum inspections, sleep apnoea detection, haemoglobin testing, vital signs like blood pressure, oxygen concentration in the blood, these are all becoming quickly and inexpensively available for consumers via their phones,” Topol says.
“This is a huge step beyond the early precursor of activity tracker and step counting. The smartphone is becoming the central hub for medicine.”
Start-ups across the globe have already begun to build portable body scanners, including smartphone stethoscopes, ophthalmoscopes to examine eyes and otoscopes to look within ears.
The NHS plans to test some of these in coming years, including the CliniCloud – a microphone that slots on to your phone to turn it into a digital stethoscope – which has been shown to exceed human doctors’ ability to diagnose chest infections in clinical trials. Another they will test is the Cordio app, which analyses change of tone in the user’s voice to predict if their congestive heart failure is getting worse.
The health service will also provide seed funding to companies such as Cupris Health, a British start-up that makes a clinical-grade otoscope that plugs into a smartphone and takes high- resolution images of the inside of your ear. As a parent with a sick child, you could do this in your bedroom at midnight. The high-res images from the mobile otoscope are sent to the cloud, where a nurse, doctor or, ultimately, an algorithm can remotely diagnose any infections.
Other devices, such as mobile ultrasound scanner Lumify, made by Philips, delve even deeper into the human body. Last January, Topol tweeted images of a head-to-toe smartphone ultrasound he performed on himself using the device – every organ from his thyroid, gall bladder and aorta, to his kidney, spleen and liver was imaged exquisitely via his phone.
“It’s $199 per month for unlimited use, compared to a $350,000 ultrasound machine in a hospital,” he says. “To me that’s revolutionary. I’d use it for every patient in my clinic.”
Doctors scan your body to look for irregularities but they rely on pathologists in the lab to accurately diagnose any infection. There, body fluids such as blood, urine or spit are tested for lurking microbes or unexpected metabolites or chemicals wreaking havoc in your body. Now companies are miniaturising these tests to create mobile pathology labs.
At the Drexel Women’s Care Centre in Philadelphia, 900 women being tested for sexually transmitted diseases have been given instant results – via their iPhones. The smartphone kit, designed by local start-up Biomeme, is being trialled as an alternative to lab-based tests for chlamydia, gonorrhoea and trichomoniasis that usually take several days to return results.
Although they are curable, sexually transmitted diseases often hide undetected for months at a time, causing serious long-term health consequences such as infertility. Any delay in diagnosis could irreversibly damage patients’ reproductive health, so a point-of-care test – one that gives the patient instant results – would be invaluable for millions.
Four-year-old Biomeme performs the same lab test via your iPhone. The plug-and-play device requires a drop of sample fluid, such as urine, and detects the genetic material of interest using a DNA amplification technique powered by its own computer chip as well as the iPhone’s WiFi, cellular signal and two cameras. It displays the results in an app.
Over nine months, the clinic has tested more than 900 urine samples using Biomeme’s smartphone kit and compared it to results from conventional methods. It found the mobile test was equal in accuracy but far easier to use. “Biomeme’s DNA isolation kit is much faster and does not require extra equipment,” the clinic’s doctors said in a presentation of the research.
“An entire test from sample to answer could be performed in approximately one hour. Thus the technology is ideal for point-of-care testing.”
The $500 kit is now being used by more than 100 public, private and academic organisations to test for the presence of everything from Ebola, Zika virus and malaria to canine distemper virus in Siberian tigers in New York’s Bronx Zoo. Each unique test costs $5 to design. The company turned cash-flow positive in 2016, making about $5m in revenues, but does not yet turn a profit.
“Our goal is how do you empower lay users to have lab capabilities anywhere? It is pretty much a doctor in your pocket,” says Max Perelman, one of the co-founders of Biomeme. “I was frustrated by the healthcare industry and services I was getting as a patient. As a non-biologist, I can use this to test myself, my children, my food and my home. I can go in my backyard and test the ticks.”
Companies such as Biomeme are radically altering the need for medical laboratories. By miniaturising diagnostic tests, they are creating mobile labs that can accurately and cheaply diagnose disease anywhere with a wireless connection.
Start-ups, including Biomeme, are targeting physicians in the field, particularly in swathes of Asia and Africa including India, Zambia, Rwanda and Kenya, where lab facilities are overburdened, and where the strict regulatory approvals demanded in western countries are not required.
But ultimately, Perelman says, Biomeme hopes to sell a consumer device that patients in the US, UK and Europe can use to test themselves, similar to a home pregnancy test. “We want complicated lab testing to be accessible to anyone,” he says.
Solutions such as Biomeme will not work for the most complex infections, when you don’t know what microbe to test for. For instance, sepsis or blood poisoning could be caused by a variety of pathogens, so the only way to catch the culprit is to sequence the DNA in a sample and look for unexpected genetic material.
Sepsis kills about 400,000 people in the US and Europe every year, mostly because it takes days to figure out what is causing it, and which antibiotics the pathogen may be resistant to.
“Right now if you developed sepsis, every hour of delay increases your chance of death by 10 per cent. Rapid treatment is the key to survival,” says Justin O’Grady, an infectious-disease microbiologist at the University of East Anglia. “If I can get a DNA result to a clinician within eight hours, that’s before they apply a second dose of antibiotics. Then they can give a personalised antibiotic for that patient and their infection.”
To speed up this process exponentially, O’Grady is testing a device called the MinION – a USB-stick-powered device that can sequence DNA in real time. Designed by Oxford Nanopore Technologies, based in the UK city of its name, the MinION uses a patented method to miniaturise and automate DNA sequencing. It is currently being used by 3,500 labs – such as O’Grady’s – to diagnose real-time outbreaks of diseases such as Ebola and pneumonia, and has even been carried on the International Space Station to sequence extraterrestrial microbes.
To scan your DNA sequence, you plug the MinION into a computer and use the company’s own analysis platform to spot the presence of known pathogens or other abnormalities. “It is still a lab tool and is not approved for diagnosis, but we tested it out on a group of children in Zambia with suspected pneumonia and found that it was able to detect the pathogen instantly,” says O’Grady.
Later this year, Oxford Nanopore will launch a smartphone-based version of its device called the SmidgeION – a little plug-in module at the bottom of an iPhone that can sequence DNA in any sample, such as spit. “This device will enable people at home to self-quantify their DNA,” says Clive Brown, chief technology officer of Oxford Nanopore. “It is the most sensitive and comprehensive way to profile what is happening in your body.”
A smartphone DNA sequencer will not only make the technology portable for physicians or lab assistants, but could help to democratise DNA-based medicine. These tests are not just useful for infectious diseases – they could be used to diagnose cancers, organ decay, heart failure and genetic disabilities like cystic fibrosis. “In the future, you would not need the clinician. A patient can potentially use it to diagnose themselves,” O’Grady says.
While mobile healthcare marches into the mainstream, serious downsides have yet to be addressed, namely privacy and efficacy. Our health records harbour some of the most sensitive details about us, from alcohol or drug abuse to sexually transmitted diseases or details of abortions – things we may never want to reveal to employers, friends or even family members.
More importantly, this dataset is permanent. It can’t be changed, like a password or credit-card number can. “It’s quite scary if you think about it. Privacy, encryption and security at this point is non-negotiable,” says Arm’s Karthik Ranjan.
As our phones become medical repositories, their security becomes even more crucial. “As a patient, you have no idea how many clouds your data has traversed through, how many times it has changed hands. What are the security policies on those various cloud servers through which your data is transiting?” Ranjan asks.
Already, medical data stores have been a prime target for cybercriminals: in 2015, US health insurers Anthem and Premera Blue Cross were hacked, compromising the medical records of more than 90 million patients. A survey that year by KPMG found 81 per cent of healthcare organisations admitted their systems had been attacked in the preceding two years.
Some security experts say that criminals are not the only ones we need to worry about – government and commercial bodies are just as untrustworthy. “Whether or not health data is encrypted from smartphone to surgery is almost completely irrelevant. Once you have digitised vast amounts of personal health information, it becomes a resource that lots of people lobby hard for access to,” says Ross Anderson, professor of security engineering at the University of Cambridge.
“We already know officials can sell on the records without your knowledge or consent on an industrial scale to drug companies and others who will abuse them.”
To prevent this, physicians and technologists recommend decentralising large health databases and allowing each patient full control over their data; that way, the risk of accessing a single enormous trove of data is reduced, and patients alone decide who can legitimately see their records: doctors, researchers or specified data custodians such as a family member.
Digital health enthusiasts also warn that each new digital tool and app must be rigorously validated and peer-reviewed, to avoid fraud – or even death.
Last year, Silicon Valley start-up Theranos, which claimed to have invented an effective finger-prick blood-test kit and had eminent personalities such as Henry Kissinger on its board, was discredited. After being outed in the press for issuing false test results, its laboratory licence was revoked, and it remains under criminal investigation by federal prosecutors for alleged deception.
“Too much in this space is unregulated, it is a jungle,” Topol says. “We need to be able to test not just the hardware and apps, but the algorithms too. If someone overdoses with insulin because of a wrong recommendation, that can be really dangerous.”
As long as digital inventions are trialled transparently, they could make healthcare enormously cheaper and more accessible to the masses. Babylon’s chief executive and founder, Ali Parsa, says it costs Babylon 80 per cent less per hour than it costs the NHS to provide medical care to a single patient, mostly because of AI assistance. In countries – including the US – where healthcare isn’t free, this could revolutionise the population’s access to medical care.
Over the next three months, NHS 111 – the non-emergency medical phone line – will trial Babylon’s app in north-central London, including areas such as Camden and Islington. At present, the team serves roughly 1.2 million people, each of whom cost the NHS £15 every time they call 111. The staff’s job is to refer to a booklet of symptoms when you call, and tell you what to do next. None are trained medical professionals.
The Babylon trial, which will require people to type symptoms into an app, will be the first government deployment of healthcare AI at this scale.
“The difference in cost isn’t a few pence, it’s 100 per cent. We can sell that to the NHS at whatever price, and they’re still saving a fortune,” Parsa says. “More importantly, we are more accurate. And our experience with the patient takes 1 minute 12 seconds on average; theirs takes 15 minutes.”
Babylon is not the only company using AI to augment human physicians’ capabilities. Hundreds of organisations are trying to apply machine learning – techniques that can sift through gargantuan data stores and pick out hidden patterns – to the diagnosis of complex illnesses.
Watson, IBM’s supercomputer, helps doctors target cancer treatments to individual patients by crunching up to two million patient records and clinical trials, 600,000 medical reports, and two million medical journal pages. In London, Google-owned DeepMind is working to speed up diagnosis of eye conditions from digital scans at the world-renowned Moorfields Eye Hospital. It is also helping to target radiotherapy treatment by analysing body scans of head and neck cancer patients at University College London Hospital.
As we begin to rely more heavily on our smartphones for diagnostics, disease management and treatment, our need for doctors will evolve. Most technologists agree that physicians will not be phased out any time soon – human and digital doctors will each perform their own niche function better than either could alone.
Many people – particularly the elderly and mentally ill – have emotional needs that exacerbate physical illness, and others have deep-seated cultural attachments to doctors. McNeil says: “That’s why the humanistic piece will never disappear.”
But although we will continue to seek out physicians, it will not necessarily be because of their superior clinical skills. “If what you need is to solve a specific clinical problem, a diagnosis, then we can diagnose you better, faster, cheaper than a human doctor can,” Parsa says, with a wry smile. “Five years from now, technologically I do not believe you will have any need to see a human doctor for diagnosis… there is no scientific reason.”
Last week Diriba Kuma, mayor of Addis Abeba, was in attendance for the...
Despite widespread concerns over foreign exchange crunch in the economy...
There could perhaps be no political party in the wo...
With recent events in the country such as mild earthquakes, the need to...
Though Ethiopia is the first African country to make modern laws, inclu...
In the 16th Century, Polish astronomer Nicolaus Copernicus published a...
A mountain of garbage that had accumulated in "koshe" gave way to a lan...