There is a reason for headlines like these — voice tech is finally blossoming. Voice follows many of the same precepts as machine learning ie the innovation is hardly in the use case, it’s in the fact that it now possible in an unprecedented way.
Why Now? Three main factors:
- Data — a lot more is available, much easier to collect, much easier to share
- Compute — much cheaper and much faster to process the data to generate insights
- Culture — deeper acceptance of using it among consumers and enterprise (providers / payors included)
Consider how voice is becoming a dominant form of UI just in people’s homes. Very fast.
What is perhaps more impressive is that voice seems to be getting adopted quicker than any other major technology in human history.
Health is an area ripe for voicetech, whether it be medication compliance, appointment reminders, diagnosing conditions as diverse as depression or lung cancer, executing tasks that require mobility among others. No wonder there is an explosion of startups in this space — the diagram below comes from an Aug 2018 article from Mobile Health that goes deeper into the 37 companies listed.
What is missing? A lot. Below are three main problems that entrepreneurs and venture capitalists alike need to solve before voice can deliver on its promise:
- Comprehensiveness — how much data should we sample to get high enough precision and accuracy (or you can express this in terms of sensitivity/specificity or false positives/negatives)?
- Privacy — is our default of anonymized and aggregated enough?
- Efficacy — can voice remind or diagnose cheaply enough or in what is otherwise difficult?