AsianScientist (Jul. 18, 2016) – The doctor instructed me to insert my arm into a blood pressure cuff. My blood pressure flashed on the screen. She beeped and whirred while she took some other measurements—height, weight, body temperature—then asked me to connect my mobile phone. I did as she asked and waited.
After combining my data with records from thousands of other patients, she printed out a single piece of paper. On it was my diagnosis. All this really happened—except it wasn’t a human doctor who saw me, but her robotic counterpart.
General practitioners (GPs) now routinely rely on automated systems in patient waiting rooms, which perform basic health tests prior to an appointment with a doctor. Is this just the beginning?
Diagnosing under pressure
Across the world, healthcare is in crisis, with hospitals experiencing shrinking budgets and a severe bed crunch. Research has shown that most visits to the GP last just seven minutes. The challenge is to describe your symptoms in under 18 seconds—on average, that’s how long you get before the doctor interrupts.
Then there’s the information overload. Doctors are increasingly faced with too much information on an unprecedented scale, from patient records to new scientific findings and ever-changing treatment guidelines. There are the blood tests, X-rays, ultrasound scans, dental examinations, blood pressure checks and electrocardiograms, in addition to the mountains of data generated by the latest gadgets and smartphones.
Perhaps it’s no surprise, then, that every day, thousands of doctors get it wrong. In the US alone there are 12 million diagnostic errors each year. In India, the misdiagnosis of malaria and tuberculosis has led to vast underestimates of prevalence rates. Could big data solve both problems in one go?
As of 2013, there was 4.4 trillion gigabytes of data in the world. That’s equivalent to six trillion CDs, or everyone on the planet sending nearly 800 emails per day for the next 1,000 years. This data deluge is already being used by businesses to predict consumer behavior and sell products. Last year, one US retailer guessed a teenager was pregnant before her father did.
But it could also save lives. With data on the habits and health of millions of people, it’s possible to draw links between drugs and side effects, symptoms and diseases—or even prevent medical emergencies before they happen. There’s just one problem: analyzing so much data requires serious brainpower.
China’s brainy computers
Chinese biotech giant BGI, based on the outskirts of Shenzhen, may have the answer. It’s the world’s largest genetic research center, with more than 178 state-of-the-art genetic sequencing machines, analyzing the equivalent of thousands of human genomes every day. As of 2010, the single facility was estimated to host more sequencing capacity than the entire United States. The company aims to sequence the genomes of a million people—one thousand million million base pairs in total—and make them available digitally.
To find patterns in this formidable dataset, BGI has teamed up with the National Supercomputing Center in Tianjin to create the Tianhe-BGI Bioinformatics and Computing Joint Laboratory. It is home to Tianhe-1A, the second-fastest supercomputer in the world.
With their digital mastermind on board—capable of 2.57 quadrillion calculations per second—BGI hopes to bring precision medicine to the masses. It works like this: companies such as BGI combine vast banks of genome data with patient health records and discover, for example, that those for whom a drug is effective always have a certain mutation. From then on, patients can be tested to avoid ineffective prescriptions and unnecessary side effects.
It’s already happening. Lung cancer patients are now routinely tested for a mutation in EGFR—a common mutation in Asian populations that indicates whether the anticancer drug erlotinib (trade name Tarceva) will make any difference.
And big data can have a big impact on emergency medicine, too. It all began with a startling statistic: despite the expertise and equipment on hand, just 17 percent of those who have a heart attack in a hospital survive. What if we could predict a ‘code blue’ in advance, and intervene? NEXT PAGE >>>