Crowdsourcing Medical Advice: Israeli Startup Steps in When Doctors Don't Have the Answers

StuffThatWorks platform employs the same model 
as Waze does to help people cope with chronic diseases by learning from each other

Send in e-mailSend in e-mail
Send in e-mailSend in e-mail
Yael Elish and Ron Held in the Negev.
Yael Elish and Ron Held in the Negev.Credit: Ofer Vaknin
Amitai Ziv
Amitai Ziv

Fifteen years ago, Yael Elish’s daughter, who was then seven, was suffering from a chronic ailment. “We saw several doctors and alternative therapists, but nothing worked. It wasn’t life-threatening, but every chronic condition is a nightmare for the family and the sick person,” she says.

The Elish family is entrepreneurial by nature, and they decided to do something.

“For years I scoured the internet every night, because I assumed there must be something that would help her. After about two years of searching, I came across a blog that described the exact same problem we had and explained how it was being treated. After that, I easily found a place in Israel offering the same treatment. We tried it, and after three weeks the problem was resolved. Because it’s a chronic illness, I can’t say she is cured – the professional term is remission, like a dormant volcano.”

A few years later, Elish came to the aid of a friend whose 5-year-old son was suffering from a chronic illness. Investigating over the internet, Elish discovered treatments that the doctors in Israel weren’t familiar with. She connected the friend with an overseas doctor who was prepared to work with an Israeli colleague.

“That was just before they were going to begin an aggressive treatment … and only because the doctor here had hardly ever seen a similar case. I realized that there are good treatments available out there, but for all kinds of reasons they don’t reach the patient: Either the doctor doesn’t know about the treatment, or you’ve got the wrong doctor, or the health maintenance organization doesn’t provide the treatment. Sometimes it happens that the doctor understands medicine but not surgery, or he doesn’t understand alternative medicine,” says Elish.

As the former head of product for Waze, the navigation software app that was acquired by Google, Elish had been part of one of the greatest Israeli high-tech success stories. She started thinking about how to use the same crowd wisdom that Waze is built on and how to apply it to chronic illness by helping sufferers find treatments doctors sometimes miss.

She started small. “I said, ‘Let’s collect the information systematically.’ I did a kind of test. I went into a chronic illness Facebook group and created a 20-item questionnaire, something very basic in Google format. The questions were: Who are you? What treatments have you tried so far? Was the treatment effective? What were the attacks like before and after?

“I started to analyze the information and draw basic conclusions – for example: 60% experienced whatever treatment they undertook as successful. I started to release graphs showing the information I’d collected. People in my circle said, ‘But what will happen if all kinds of people who don’t have a clue contaminate the data so it isn’t accurate?’ But people took the questions very seriously and cooperated.

“Even when I lengthened the questionnaire and added new questions, people didn’t stop. They answered, and they were also able to answer better than the medical world tends to think. There’s a certain disdain of patients and their stories. They’re considered anecdotes. But it’s people who know how to describe their illness best.”

From that, Elish created the startup StuffThatWorks two years ago. Its aim is to transform patients’ anecdotes into a coherent picture of an illness and the treatments that exist for it – precise and up-to-date information based on a broad sample.

StuffThatWorks staff at the company office in the Negev. Credit: Ofer Vaknin

Elish says there are some 10,000 recognized chronic illnesses and about 500 treatments for each of them, ranging from medications to lifestyle changes to supplements. Most of the research is directed toward the most common illnesses, which number between 20 and 40.

Are you saying the medical profession doesn’t monitor the success of a treatment or a medicine?

A great deal of money is invested in pharma, but the pharmaceutical companies conduct most of their clinical studies before the treatment is put on the market. Once it’s out and you can get what’s known as real-world evidence, little research is done. The reason is cost. One patient in a study of that kind costs the pharmaceutical company $10,000 and represents a single case, so large numbers are needed to measure effectiveness. When you get down to the subgroups of an illness and the number of possible treatments, you realize that because of the insane costs, the only way to achieve mapping is by mobilizing people to do it together.”

Elish drew on her Waze experience: Waze mapped the world from scratch; as people traveled, they drew a map of the world’s roads. “It’s classic crowdsourcing – take a big problem that’s dealt with ponderously today, with plenty of survey companies in the middle and manual data feed and the like, and provide a crowdsourcing-based solution,” she says.

Elish founded StuffThatWorks together with Yossi Synett and Ron Held. Synett, a physicist by training, is the company’s head of data science; Held, the chief technology officer, is a mathematician with 10 years’ experience in a secret Israeli army unit. Elish, who has a degree in international relations, is highly respected in the Israeli tech industry. As to the logic behind the startup, Elish notes that 8% of all internet searches are health-related and half of all Facebook groups deal with the topic. “But the information is scattered all over the place and it’s hard to find your way,” she says.

Held elaborates: “The dialogue on the web is strongly tilted toward the most vociferous writers, those who express their opinion vigorously. There’s also no specific reference to the patient’s background.” That’s the problem the company is trying to address.

Elish, 52, joined Waze in 2008, after her previous startup, eSnips, was sold. Waze was a tiny startup that compromised the founders and five developers. Elish was in charge of user experience and branding. She says: “We made decisions that were revolutionary at the time, such as free text searches, in contrast to the usual format that existed then in the world of mapping, of click on city, street, street number. We put an emphasis on the ‘cuteness’ of the product’s look and its game quality. That proved to be critical.”

The Waze team debated whether a “cute” product would be perceived as credible. It turned out that it could, and she’s now out to replicate that style with STW.

“We have a wonderful designer and illustrator, Maya Shleifer, who’s illustrated several children’s books and an interface that’s pleasing to the eye. Our name, StuffThatWorks, is ‘not right’ in any branding sense, but it represents people, information that comes from people. There’s some sort of apprehension that if you’re not called Med-something or Health-something, and your company logo doesn’t have a serpent, you’re not serious. But we say the product will be cute and it will be serious.”

Until a couple of weeks ago, STW was operating under the radar, but the whispers about it in the tech industry grew loud. In July 2018, the company raised seed capital of $3.5 million from leading VC funds 83North and Bessemer Venture Partners, for which investing in a new startup is unusual. A year later, Ofek Ventures put in another $5 million. On July 23 the site launched worldwide, for now only in English.

The platform has more than 180,000 active users, meaning that they are not only registered but also share information. Another thousand join every day. There are currently 105 chronic illnesses mapped on the platform. “We call them research communities,” Elish says. “Everyone can open a new community and invite people to fill in a questionnaire and share. Once there are 30 active participants in a community, it opens officially. At the moment we have 200 of them in waiting.”

Participants write posts about their experiences and add questions. The system informs users about others who have a similar condition, so they can communicate privately. STW utilizes natural language processing to obtain an in-depth understanding of what people are saying. “There’s a difference between ‘my head hurts’ and ‘my head feels like it’s going to explode,’” Held explains. “A sentence like that can be compared to the way a person expresses himself in general or describes other symptoms. You can find biases in the language, for example, for or against alternative treatments, even detect deceptions, like someone who’s there to sell something.”

Like every crowd wisdom startup, STW also faces the hurdle of harnessing people for the community before the product adds value, like the first users of Waze, who helped map Israel before Waze could provide navigation. “You have to persuade people to invest in the vision, so the user feels it’s heading to a cool place and says, ‘I want to be part of that,’” she says. “There will always be someone with the vision to get a community off the ground and induce people to respond. We view those people as part of the company.”

Held notes that STW employs a wider definition of health than doctors conventionally do: “Medicine takes note mainly of clinical indices, so sometimes it will deem you healthy, even though you’re depressed, can’t bring yourself to leave the house or aren’t functioning sexually. The cancer didn’t recur? Then that’s that. With chronic illnesses the most important measure is quality of life, along with the clinical measures.

“Sometimes we don’t understand what the patient wants to solve. Some of the female patients in a particular group were upset because they put on weight and others, about possible damage to fertility. If an infant suffers from reflux, as far as the doctors are concerned there’s no illness as long as the baby is gaining weight; but from the parents’ viewpoint the child isn’t able to sleep and it’s impossible to go anywhere in the car. Our data respond to a far larger range of patients’ wishes than the world of traditional medicine.”

I can imagine the doctors are saying that what you’re doing isn’t serious or that it’s fake news that will adversely affect public health.

Every doctor we’ve shown it to thinks it’s great. ... They love us, because they readily admit that this is a blind spot for them. For them, every small research study takes half a year – we already have 10 million data points.”

The company is only just starting to see success. “If you look at depression,” Synett says, “you’ll see that the treatment that’s ranked 25th in use is in second place in the effectiveness ranking. And that treatment, by the way, was developed for a different psychiatric condition. Apparently a psychiatrist prescribed it for a patient who suffers from both, saw that it helps and started to try it on other patients.”

According to Held, there have cases in which STW identified a particular treatment as effective shortly after it came on the market, a year before medical studies began corroborating the evidence. “Crowd wisdom can be ahead of research, indicate new treatments as effective relatively early, even help set a direction for future research,” he says.

For the users, the information is provided free of charge and will stay that way. Most information on the site can be accessed via Google, without registering. Academic bodies and scientists will also get access to the raw data for free. But the company intends to commercialize the information (in the aggregate and only statistically, to ensure privacy) and sell it to clinics, hospitals, insurers and pharmaceutical companies so they can recommend better treatment and save insurance costs, for example.

When the coronavirus broke out, six STW employees and their families, as well as a few singles, came to live on a farm deep in the Negev desert; other employees would come and go as they pleased, Elish says. “We were on cloud nine. In the morning, the children went to the goats and we started to work. ... It was good for the adults, too, without the pressure of having to take the kids to preschool with traffic jams. It was a wild period for the whole company, not only for those who worked from the farm.”

STW is thinking about a future between city and desert. “Now we’re thinking seriously about some sort of hybrid model between the center and the desert, and we already have one employee who moved to the region,” says Elish. “Before the pandemic, we were going to sign for a new office space at five times the cost of the present office in Tel Aviv’s Old North. That won’t happen.”

Click the alert icon to follow topics: