"Alexa, Am I pregnant?": A content analysis of a virtual assistant's responses to prenatal health questions during the COVID-19 pandemic.

Abstract

To elucidate whether Amazon's virtual assistant, Alexa, provides evidence-based support as a supplement to provider-facilitated prenatal care, during the COVID-19 pandemic. Using a conceptual content analysis approach, a query of 40 questions, relating to all phases of pregnancy, was collected from Alexa by two independent investigators, using two unique devices, over a one-week period between May 20, 2020 and May 27, 2020. Alexa's responses were matched to the evidence-based content from the American College of Obstetricians and Gynecologists (ACOG) and reviewed by a Certified Nurse Midwife for completeness and currency. Of the 40 questions asked of Alexa, it was unable to answer 14 questions (35%). A total of 21 out of the 40 responses (52%) were not evidence-based and three COVID-specific questions (about 1%) were answered incorrectly or insufficiently. Four questions (10%) were answered accurately. Alexa was largely unable to provide evidence-based answers to commonly asked pregnancy questions and, in many cases, supplied inaccurate, incomplete, or completely unrelated answers that could further confuse health consumers. Ensuring that mobile health (mhealth) tools, such as Amazon Alexa, are evidence-based and credible in answering common prenatal questions has important implications for this pandemic and future consumer needs.