Dr. Kira Radinsky wants to change the world. But unlike your average 32-year-old, she may well be on the path to doing just that.
“There’s a wide change that the world is going through – especially with all the data we’ve been accumulating,” she tells Haaretz by phone. In her case, as a data scientist whose research predicts future events, that information is currently millions of health records that provide insight on large populations and suggest personalized treatments for multitudes of patients.
“I’m interested in how we can improve the hypothesis and [make] medical breakthroughs in hours instead of years,” she explains.
By her mid-20s, Radinsky – who was born in Ukraine but immigrated to Israel at age 4 – had already earned a PhD in computer science, predicted riots in Turkey and Syria and a cholera outbreak in Cuba, co-founded a company that was later sold to eBay for $40 million and won numerous awards.
Today, she is both eBay’s director of data science in Israel and visiting professor at her alma mater, the Technion – Israel Institute of Technology, Haifa, where she started studying for her first degree at 15.
Radinsky’s passion is creating predictive algorithms that enable early detection of globally impactful events, such as the spread of diseases. She spots clues in the past and present, which allow her to literally predict the future and potentially save lives.
The research she’s currently leading at the Technion draws on records gathered over the past 20 years in Israel’s Electronic Medical (or Health) Record system, identifying drugs that people take to treat certain diseases but which actually also help with other ailments they have.
“In pharma, developing a new drug every year is twice as expensive and takes twice the time – in other words, we’re running out of time,” Radinsky says. So she’s trying to identify drugs that are already FDA-approved and which treat more than just the intended illness.
She’s already getting positive results. Her research includes a partnership with Israeli health maintenance organization Maccabi and she has found hundreds of different drugs – or combinations of known drugs – that can be repurposed to improve treatments.
For example, they’ve discovered a connection between hypertension (high blood pressure) and proton pump inhibitors (a group of drugs targeted at reducing stomach acid production). It normally takes six to nine months to balance the symptoms of hypertension, but Radinsky says her system cuts that to fewer than three months.
Knowing the range of diseases of family members and the drugs they’re taking also helps Radinsky predict which illnesses their relatives are likely to have and what drugs they should be taking.
“We’re actually building a national AI system that looks at millions of people and gives advice to doctors for better treatment” – especially those who are seeing patients for the first time, Radinsky explains.
Even more fascinating is her deep learning and retrospective study, which creates a system that generates molecules for specific targets. If that sounds like science fiction to you, you’re not alone, but Radinsky provides a simplified example: Using only data that was available in the 1930s, her system generated from scratch the molecule for a tuberculosis drug that only became available in the ’50s. “The number one drug that the World Health Organization requires every country to have – and our system would have discovered it 20 years before it came to the market,” she says.
Her main goal now is to bring her passion to the world. On July 12, Radinsky was appointed to the 20-member United Nations High-level Panel on Digital Cooperation, launched by UN Secretary-General António Guterres.
The panel is co-chaired by Melinda Gates, who also co-heads the Bill & Melinda Gates Foundation, and Jack Ma, the billionaire executive chairman of tech conglomerate Alibaba. It will assemble for the first gathering of its nine-month run in September.
Guterres has asked the panel members – including a United Arab Emirates government minister, a Google vice president and a Chilean professor at Harvard – to identify policy, research and information gaps in the digital space, and propose ways to strengthen international cooperation.
“If all people, especially the poorest and most vulnerable, have equal access to digital technology, they will use it to improve life for themselves and their families, and raise their voices in conversations about what the future holds,” Gates said in the launch statement.
For Radinsky, this is an opportunity to share her discoveries on the global stage.
“If each and every one of us is going to continue doing their research in silos, I don’t think we’re going to scale,” she says. Alongside her new colleagues, she hopes to find a way to unite all the data the world has – whether it’s a worldwide genomic archive or a global EMR system.
Exchanging data in this day and age also poses a serious challenge. Radinsky, who is a proponent of large-scale data sharing, is fully aware of the existing risks and the significance of making beneficial use of the data without hurting people’s privacy. “There are a lot of mathematical models of how to reach this, but we need to agree on some protocol that everybody accepts,” she says – an end goal she hopes to achieve throughout the panel discussions.
In this post-Cambridge Analytica era – referring to the now-defunct data-mining firm that used personal information extracted from over 50 million Facebook profiles without permission to help influence the 2016 U.S. presidential election – data scientists emphasize the difference between the need to regulate data acquisition versus regulating applications.
“Personally, I’m a big fan of not regulating data acquisition but rather, the applications of it,” Radinsky says. “Cases like the one Cambridge Analytica created are not about Facebook collecting our data and using it for its own benefit; it’s about companies like Cambridge Analytica making applications that eventually don’t provide value for the customer.”
Radinsky advocates differential privacy, a new form of cybersecurity that prevents private data loss in cases where masses of data is used, mainly by introducing randomness to the data analysis results.
Another challenge facing Radinsky and her peers is data bias. If, for instance, a police precinct is continuously arresting individuals of certain backgrounds and ethnicities, predictive algorithms will continue the bias because their forecasts are based on patterns. “We need to think about ways of how to ‘debias’ the data, how we actually require fairness [from] the data for different types of applications as well,” Radinsky says.
As an entrepreneur and scientist, she also recognizes her role as a female leader in her specialized fields.
“Getting more women into this field of science, research and engineering is so beneficial,” Radinsky says. She points out that certain diseases only affect women, for example, and only a gender-diverse group of researchers will be able to deeply understand and treat them.
As a working mother, pregnant with her second child, Radinsky hopes her work will serve as a role model for young professionals who want to go into science and engineering. “I do want to promote the younger generation to do this transformational work and not be afraid, whether they’re men or women,” she says.
Want to enjoy 'Zen' reading - with no ads and just the article? Subscribe todaySubscribe now