IDAvatars, Working With Watson, Seeks to Answer Patients’ Questions
(Page 2 of 2)
having developed similar capabilities in Watson. The supercomputer’s ability to listen, process, and quickly respond was on display when it took on human opponents in the game show “Jeopardy!”
Part of the reason iDAvatars uses proprietary technologies in tandem with ones developed by IBM is that Watson is continually becoming more knowledgeable and powerful. Daroga says when his team started to map out how to integrate with the supercomputer two years ago, IBM only offered a single application programming interface to its partners. Now more than 30 are available, he says.
Another improvement Big Blue has made is giving sets of questions a more robust back end through a “natural language classifier” tagging technique. The way it works, describes Daroga, is users rate answers they receive from Watson, which factors in the feedback when someone asks a similar question in the future.
“Over time, Watson picks up nuances,” Daroga says. “It learns that the question really being asked is a little bit different from what we think the person is asking us.”
The company’s other customers include German pharma Bayer, and Intel (NASDAQ: INTC), which uses motion-sensing technology developed by iDAvatars in its RealSense 3D Camera. Daroga says PC manufacturers like Lenovo, Dell, and HP (NYSE: HPQ) are working to integrate the Intel cameras into their laptops, one of which he used to run this reporter through a quick exercise with Sophie.
She instructed me to raise my hand and switch between splaying my fingers and clenching them into a fist as many times as possible in 15 seconds. The camera, which can also track range of motion, counted my reps correctly and the results suggested I’m slightly faster with my dominant right hand than I am with my left. Daroga says patients who have suffered nerve damage might supplement their physical therapy regimens by interacting with Sophie. One advantage of high-tech rehabilitation is patients can monitor their progress without having to log anything themselves, he says.
IDAvatars, which Daroga says has about 20 employees worldwide, has raised a total of about $3.5 million from investors. The company had sought to raise $5 million or more in a Series A round, but it has started to make enough sales that it’s now bootstrapping. Daroga says he expects the company to be profitable in 2015 on projected revenues of $1 million.
For now, the company is focused on signing up additional customers, including South Korean electronics giant Samsung. Daroga says he’s met with executives from the company to explore incorporating iDAvatars-created characters into Samsung’s S Health wellness app.
While there appears to be significant growth potential, developing avatars who speak other languages could be challenging. The voice-to-text technology that iDAvatars developed for its characters cannot be easily adapted to other languages, Daroga says, though he notes that English is the predominant language for medical terms. He says developers could quickly make headway on simple questions and answers, and other low-hanging fruit. However, a significant investment in translation technology would be required to make avatars designed by his company—whose motto is “The Art of Empathy”—multilingual.