Big data emphasis underpins Teva's plans for smart inhaler, clinical trial innovations

At the MEDinIsrael conference in Tel Aviv, Director of Analytics and Big Data Dr. Lena Granovsky talked about technology, big data, and analytics' growing role at the company.
By Jonah Comstock
02:41 pm
Share

At MEDinIsrael 2019 this week, a conference in Tel Aviv sponsored by the Israeli Export Institute, Dr. Lena Granovsky, director of analytics and big data at Teva Pharmaceuticals, spoke on and off stage about the work the top Israeli pharma company has been doing in digital health. These topics included the company's recently FDA-cleared smart inhaler ProAir Digihaler, as well as more broad discussions about the potential for machine learning and digital health in clinical trials and drug development.

“There is a huge unmet need,” Granovsky said during a press briefing. “Asthma and COPD patients, they [often] don’t know how to use their inhaler properly. Especially when they’re in a rush, they’re not sure their technique is correct. All the information is in the patient application it has all the reports about the patient, so the physician can [review the data and say] ‘Your technique wasn’t the proper technique’”

Today adherence, tomorrow prevention?

When Teva bought Cambridge, Massachusetts-based Gecko Health back in 2015, it was a significant moment for digital health — at the time, it was fairly novel for a pharma company, especially a major one, to buy a digital health company.

Yet the technology itself, a sensor-laden inhaler that tracks adherence, was less novel even in 2015, with s smaller pharma called Opko Health having bought a similar startup by the name of Inspiro medical the previous year. And that's not to mention Propeller Health, which had already been developing something similar in partnership with a number of large pharmas.

But it’s only now that Teva’s endgame from that acquisition is being revealed, and as the company gets close to its planned 2020 rollout of its connected inhaler, it’s clear that the technology has evolved quite a bit over the last four years behind the walls of Teva’s R&D department.

While many connected inhalers are focused only on measuring how and when they are being used, Teva’s new inhaler can also measure peak inspiratory flow rate. That’s a metric competitors don’t measure — and one that Gecko’s inhaler didn’t measure in 2015. While that data might not be as immediately valuable as adherence data, it could prove to be vital for Teva’s big data aspirations.

“We can take this data from this smart inhaler about time and number of inhalations, flow rate, about volume, and we can build an algorithm that can predict asthma exacerbation,” Granovsky said on stage at the event. “If we could predict exacerbation five or six days before it hits, the research shows that it’s very possible we could prescribe some drugs and avoid the exacerbation.”

And according to Granovsky, Gecko has catalyzed a culture change at Teva as well.

“This team kind of started this technology revolution within Teva,” she said.

AI has potential at each step on the road to drug development

ProAir Digital Inhaler is a beyond the pill, medication wraparound sort of innovation. But big data and predictive analytics have also become big priorities for other parts of the company, which see value to be gained from these technologies at every stage of the expensive drug development process.

Only about 5 percent of drug candidates become drugs, and those that don’t make the cut can fail out for a number of reasons.

“Sometimes the drug fails not because of oncology, but because of the limitation of the development process and the way we are measuring the success of the drug,” she said, noting that patients failing to accurately and promptly fill in patient diaries, for instance, can lead to inaccurate results when they fill in a diary entry after the fact from memory. “[Or] sometimes it might happen that the drug fails because it’s the wrong patient population. Maybe the drug works, but not for everybody. We can make wrong assumptions.”

Measuring the wrong endpoint or outcome is another risk. Granovsky described a hypothetical situation in which a drug might make a patient feel better, but that in turn could lead the patient to be more active, causing back pain they didn’t have when they were sedentary. But if only pain level is managed and it’s only managed at monthly intervals, the results could miss the improvement in quality of life and reflect only the rise in pain level.

For many of these bottlenecks, the right digital health platform that lets patients report in the moment via an app, or uses data analytics to intelligently segment populations, can help to counteract the many forces preventing drugs from making it to market.

“Each and every one of these things can cause a drug to fail,” Granovsky said on stage. “And if the drug fails because it doesn’t work, that’s a good thing. But if the drug fails when it actually works, but we couldn’t prove that it works, that’s something that is actually heartbreaking for me. Because there are patients that could benefit from this drug’s development and patients this drug could help.”


Editor's note: This story was reported in Tel Aviv, Israel on a trip paid for by the Israeli Export Institute, an organization funded in part by the Israeli government, which covered airfare and lodging for reporters. MobiHealthNews Editor in Chief Jonah Comstock also participated as an onstage speaker at MEDinIsrael 2019. As always, MobiHealthNews maintains its editorial independence and made no promises to the Israeli Export Institute about the content or quantity of coverage.

Share