Fined over fake-drug ads, now inside the medical industry: what has Google been doing to prepare?

Ten years ago, Google's relationship with medicine amounted to selling ads — many of them fake.

Last week, the French pharmaceutical giant Sanofi announced a joint venture with Verily, Alphabet's life-sciences company. Each side is putting in $250 million to develop tests and treatments for type 2 diabetes.

Verily started as a medical-technology team inside Google. After Google reorganized itself into Alphabet last year, it became a separate company and announced a joint venture with Johnson & Johnson to build surgical robots.

Just last month, Verily set up another joint venture, Galvani Bioelectronics, with GlaxoSmithKline, one of the world's three largest drugmakers. It will develop implantable miniature devices and drugs that treat a range of chronic diseases by changing the electrical signals between the nervous system and the body's tissues.

Google's medical news used to read like science fiction: contact lenses that monitor diabetes, technologies that extend life. You can find that sort of thing in the research labs of any number of old-line tech companies. It sounds far ahead of its time, and it stays far away from practical use.

These three joint ventures are different. Each pairs Google with a top drugmaker and hundreds of millions of dollars from both sides; each splits ownership 50:50 or 40:60; each aims at one of the big markets in medicine — diabetes, the diagnosis and treatment of nervous-system disease, and medical devices.

Google, a tech giant that takes more than 90% of its revenue from advertising, looks like it has genuinely entered the medical industry.

At first, Google was no better at medicine than Baidu

In the beginning, Google's only tie to the medical industry was search advertising. Ads for drugs, insurance and health supplements have long been a main source of its online ad revenue.

Like Baidu's, Google's medical ads were once full of false claims.

In 2003, three separate committees of the U.S. Congress questioned Google over online drug ads. Google put some safeguards in place — much as Baidu did after its own scandal broke — and the fake medical ads stayed.

Then came 2009. The Federal Bureau of Investigation ran a sting, sending a counterfeit-drug dealer to buy fake medical ads on Google. Google's ad sales team gave him a dedicated account rep, who taught him how to get around Google's own filters.

It cost Google $500 million, and the company admitted it had a medical-advertising problem. Google then rewrote its rules for ad placement and used them to review and clear out ads. Over the next year it pulled 130 million false ads.

Search a medical term on Google today and the results mark "Ad" in a prominent yellow box, so you know what you are looking at. On the whole page, that yellow label is the thing your eye finds first. Inside Google, a team of a thousand people does nothing but take down ads that break the rules.

Medical ads bring in less money now, but no fewer people come to Google looking for medical information. By Google's own count, in 2010 in the United States alone, 100 million people ran 4.6 billion such searches.

What Google did next was no different from Microsoft

Once medical ads got harder, Google tried to help doctors and hospitals move care online. In 2008 it launched Google Health, a service for managing personal medical records.

Google Wave, unveiled at Google's developer conference that same year, was used to manage electronic medical records in the United States. Then came Google Helpouts, a general advisory platform where users could put medical questions to someone over video chat — close to an online consultation.

In February 2014, Google struck a deal with the diagnostics company Quest Diagnostics under which more than 100,000 doctors on its platform would supply users with information about diagnostic tests.

All of it amounted to a tech company doing for doctors and hospitals what it already knew how to do — managing data, answering questions online. It was a hospital IT department.

None of these mild efforts led anywhere. They did not change how people see a doctor, and they did not earn Google any real money.

"After those first attempts failed, tech companies became more careful about claiming that digital solutions would transform healthcare," Bertalan Mesko, author of The Guide to the Future of Medicine (《颠覆性医疗革命》), told QDaily. "I think they have realized this is a more complicated industry than they imagined."

Google's first real effort in medicine began with an investment

In 2007, Google invested in 23andMe, the consumer gene-sequencing company Anne Wojcicki had just founded. It was Google's first medical investment.

The relationship was not an ordinary one: Wojcicki is the ex-wife of Google co-founder Sergey Brin, and the two were still married when the money went in.

The company sold a DNA test for $99. You bought the kit, spat 5 ml of saliva into it and mailed it back, and some time later received a report listing your risk of inherited disease.

At the time, gene testing was still distant and expensive for ordinary people. As the world's first cheap genetic test, 23andMe became a sensation in America; doctors and the founder went on the evening programs of the major networks to explain a product that had belonged to science fiction.

What 23andMe was doing turned out to be less simple than it first thought. In 2013 the U.S. Food and Drug Administration barred it from using test results for medical diagnosis. Two years passed before it won approval to offer gene sequencing for Bloom syndrome, a rare inherited disorder.

23andMe more or less sums up Google's later investment style: hire a lot of researchers, chase frontier technology, and run into enormous obstacles on the way to the clinic.

The investments picked up in 2009, when Google set up a separate venture fund, Google Ventures.

To start it, Google CEO Larry Page brought in the investor Bill Maris, who had done neuroscience research at Duke University and, after graduating, invested in biotechnology and health.

GV has always run independently of Google, taking $300 million to $500 million a year from the company and spending it as it likes across every field. Life sciences, healthcare and artificial intelligence are among its most important areas; two of its four core partners cover life sciences and medicine.

Since it started, GV has backed more than 350 companies, over 100 of them medical. Those bets concentrate on frontier work in genetics and cancer, or on helping the industry handle large volumes of data — the cancer big-data firm Flatiron Health and the DNA cloud database DNAnexus are typical.

Seven years on, GV is one of Silicon Valley's most successful venture funds. By number of successful exits, a CB Insights report puts it second among corporate venture arms, behind only Intel Capital.

Uber, the workplace chat software Slack and Jet.com, the online retailer Walmart bought for $3 billion, have all taken GV money.

What matters more to Google is that GV has become an authority on medical-technology investing.

This year the blood-testing startup Theranos was found to have faked its technology, and its $9 billion valuation went to zero overnight.

When The New York Times and others went back over what had gone wrong, they counted GV's two refusals to invest as one of the warning signs.

A GV investment has become a way for a new medical-technology company to be taken seriously. That brings GV itself more deals, and it also helps Google find companies worth buying. Some of those acquisitions became the core of Google's own medical teams.

Google also took up medicine in-house, in the most traditional way

Around the time Google Ventures was founded, Google began putting together a secretive lab of its own, Google X, which produced Google Glass and the self-driving car.

2007

Google invests in 23andMe; the deal is led by Google's venture-investment arm

2009

Google sets up a separate venture fund, Google Ventures (GV)

2010

Work begins on the Google X lab

2014

Google X announces a run of frontier projects: glucose-monitoring contact lenses, cancer-fighting nanoparticle robots, and Baseline Study, a project to build a genetic database of healthy humans

2015.8

Google splits and reorganizes; the Life Sciences division and the secretive-projects lab Google X both become independent subsidiaries of Alphabet

2015.12

Life Sciences is renamed Verily

2015.12

Verily forms a joint venture with Johnson & Johnson

2016.8

Verily forms a joint venture with GlaxoSmithKline

2016.9

Verily announces a joint venture with Sanofi

Jessica Mega was a well-known cardiologist at Brigham and Women's Hospital and Harvard Medical School. She says her career plan never included a job outside a hospital. Recently she joined Verily, Google's life-sciences company, as chief medical officer.

Running the genetics lab at Brigham and Women's, Mega had realized that progress in genetics depended on analyzing data at scale, so she went to friends at Google and asked them to help organize a mountain of genetic information.

That is the most common way Google works with medical institutions on research: Google supplies the computing, as other tech companies do, and runs joint projects with Stanford, Duke, Harvard and other universities.

A fair number of those projects were absorbed outright into Google.

Google X's glucose-monitoring contact lens began as a piece of research Microsoft did with the University of Washington. The research team ended up at Google X. "At a university you can take this research into the academic world, but at Google we had the chance to be funded," said Brian Otis, one of its co-founders, explaining the move.

In 2013 Google X hired Andrew Conrad, a medical scientist with a legend attached to him: he had once led work on a serum test for HIV. That work became the starting point for Baseline Study, one of Google X's major projects.

The plan is to collect the genes and other information of healthy people and use the data to draw a complete picture of a healthy human being, so that the risk of disease can be caught before it appears. As the sample base grows, the team will use the data to redefine what a "healthy human" is.

In his 19 months at Google X, Conrad built a team of more than a hundred scientists to work on cancer-fighting nanoparticle robots: a pill carrying magnetic nanoparticles that, once swallowed, enters the bloodstream and, with a matching wearable device, detects and reports data about the body as it travels.

As with the earlier university work, this Tricorder project was carried out with researchers from MIT, Stanford and Duke.

Asked by Steven Levy, author of Google: Disruptive Innovation (《Google：颠覆式创新》), why he chose Google rather than a university, Conrad said: "Anywhere else, these things might take twenty years."

The contact lens that measures blood sugar is the best known of these projects. Google X announced a partnership with Novartis to commercialize it, placing a sensor thinner than a human hair between two layers of the lens and reading blood-glucose levels in real time from tear fluid.

All of these projects have lately been reported to be moving slowly; the research may be going much slower than anyone imagined.

But as Conrad put it: "We have a million crazy ideas about what can happen inside a human being, and putting a pill inside a person is still a long journey. But we think it is a matter of a few years, not a few decades."

After Google's reorganization last August, one part of Google X, Life Sciences, was split off into a subsidiary of its own, Verily.

Verily now holds more than 150 chemists, biologists and machine-learning and data-mining specialists, with Andrew Conrad as CEO. Once it became one of Alphabet's letters, he positioned Verily as a company "moving away from traditional medical technology."

"The contact lens that measures blood sugar looks more like science fiction, but there are rational investments too — the partnerships with Johnson & Johnson and GlaxoSmithKline, and the spoon that helps Alzheimer's patients steady their tremors," Mesko said of the team Conrad leads.

Lately, Google has turned the thing it does best, artificial intelligence, on medicine

Another Alphabet subsidiary working closely with medical institutions is DeepMind, the artificial-intelligence company — the one that does Go.

It recently announced its latest project with Britain's National Health Service: working with Moorfields Eye Hospital in London to let artificial intelligence help doctors diagnose and prevent eye disease.

In a research project announced a few months ago, DeepMind also obtained access to the medical records of 1.6 million patients in London.

That project has each side covering what the other lacks. People cannot work through data this fine-grained quickly; machines can. Machines do not know what any of it means; researchers can teach them what an anomaly signifies.

This is the most valuable thing Google brings to its medical partnerships — data-processing technology and models applied to an industry that could never handle data in bulk before. Medicine has reached the point where it needs artificial intelligence, and Google happens to be one of the largest artificial-intelligence companies in the world.

Google's engineers taught AlphaGo to play Go. They are also teaching artificial intelligence to read X-rays, and to acquire the judgment a doctor needs years of study to build.

Google did not invent this. Dozens of funded artificial-intelligence startups are already trying to teach machines to recognize disease, and they are all doing much the same thing: not replacing experienced doctors, but taking the simple, repetitive part of daily diagnosis off their hands so they can spend their time where intelligence and experience matter.

Seven years, hundreds of investments, a team of more than a thousand people. Google, once no better at the medical business than Baidu, has finally begun to enter the industry.

Cover image from gv.com and The Next Web