As providers look to further integrate voice assistants into clinical care, some options are more effective at answering patients' queries than others, according to a recent study published in the Annals of Family Medicine.
For the study, Palo Alto, Calif.-based Stanford Healthcare researchers analyzed four major voice assistants' ability to understand queries from cancer patients. The four voice assistants were Amazon Alexa, Apple Siri, Google Assistant and Microsoft Cortana.
The research team used their personal smartphones for the analysis, in which they asked each voice assistant the following question, twice, for 11 cancer types: "Should I get screened for (type of) cancer?" The researchers then assessed each voice assistant's ability to understand the queries, provide accurate information through web searches and verbally convey the information.
The voice assistants' responses were compared to the U.S. Preventive Services Task Force's cancer screening guidelines; responses were considered accurate if they did not directly contradict the guidance.
Four study insights:
1. Siri, Google Assistant and Cortana understood 100 percent of the queries and consistently provided a web search and/or verbal response. Alexa was unable to understand or respond to any of the queries.
2. The top three weblinks recommended by Siri, Google Assistant and Cortana provided information that matched the task force's guidelines about 70 percent of the time.
3. For verbal response accuracy, Google Assistant aligned with the task force's guidelines 64 percent of the time, which was similar to its accuracy rate of web searches. Cortana's accuracy of 45 percent was lower than its web searches, and Siri was unable to provide a verbal response to any of the queries.
4. All voice assistants' verbal responses to queries were either unavailable or less accurate than those generated by manual web searches.