Paving the Path for Voice-Controlled AI in Pharma

When Amazon released the voice-controlled speaker Echo in 2015, artificial intelligence made what some considered a giant leap forward. Although the online retailer and innovator doesn’t publicize sales figures, it has claimed major growth and was projected to own 70% of the market this year. Alexa, as the device is commonly known, could become a $10 billion boon by 2020, according to researchers.

AI has found its place in the home. But what about in healthcare and pharmaceuticals?

A panel of tech-heavy experts at the Digital Pharma East conference in Philadelphia this week grappled with how AI voice assistants can become useful tools in the industry. Their answers hit on the technology itself, regulations, and corporate culture. But they agreed that the Echo and its peers hold great opportunity.

“It’s natural to just speak,” Vlad Castillo, co-founder of xReach.io, which focuses on voice technology for healthcare and life sciences, said. “We have the ability to open the door to that content that we have and work so hard for.”

The first thing to know, he said, is that Alexa accurately interprets commands about 95 percent of the time. That includes people with accents, wherever they are from. The impressive level of understanding bodes well for its future use in pharma, he added, though others have warned that the tech must be fail-proof in clinical settings.

But challenges remain. Ryan Billings, senior director of digital innovation for AMAG Pharmaceuticals, said voice-driven AI must learn to understand sentiment. Existing social media data collectors sort specific words by whether they are deemed positive, negative, or neutral, which does not always get at a user’s true feelings, he said.

“With voice, we can actually get that emotion factor,” Billings said, “so we can have the machines understand that and get as smart as we are and get emotional.”

This issue is scalability, Glenn Butcher, senior director of global cystic fibrosis marketing for Vertex Pharmaceuticals, said. Yet machine learning can solve that problem, he said, enabling the AI itself to overcome a barrier that has thus far baffled human brains.

Beyond the technology, pharma and healthcare must weigh how much they trust AI to carry out digital tasks—and conversations—that have real-world consequences. That is especially important in pharma, with its complex web of regulations, panelists said.

Castillo offered one simple solution: Before providing information, voice assistants can tell the patient to say the word “help” at any time for a product’s important safety information. Creative fixes like that, he said, can help healthcare and AI companies manage risks that sometimes keep institutions from embracing innovation.

To that end, businesses that opt to pursue this technology must go all in, according to the panel. Billings suggested they start small, by beefing up their social media presence, and then progress. And internal education is crucial to creating success on these new frontiers, he said. No one can wing it.

“If you’re going to be reviewing digital things,” he added, “you need to be using it in your personal life. So get an Alexa. Get on Facebook. But get comfortable with this technology because it’s happening.”