Evaluating the use of chatbot during pregnancy: A usability study

Reduced anxiety is critical for the baby’s health during the thousand days of pregnancy. In this age of false information, unreliable data can be harmful to pregnant women during prenatal and postnatal periods. This work presents the findings of a pilot study that investigated the use of chatbots to assist pregnant women during the prenatal and postnatal periods in Brazil. We conducted experiments with healthcare professionals and pregnant women, using the parallel convergent mixed method design to compare the perceptions of the two groups. We applied a quantitative/qualitative study using data collection tools available via the internet. Before responding to the survey instrument, both groups spent seven days interacting with the chatbot. Two doctors validated the questionnaires, and we used a previously validated questionnaire for pregnant women. Seven physicians’ and thirteen pregnant women who met the search criteria participated in the study. Pregnant women believe that interacting with the chatbot educated them and, their physicians will approve of their use. The most significant and positive construct was related to the chatbot’s performance expectation ([Mean 4.61][SD 0.74]). The construct that had the least positive influence on pregnant women was Facilitating Conditions ([Mean 3.30],[SD 1.24]). Pregnant women, according to physicians, benefit from clear language and comprehensive information in a variety of ways. Finally, it was established that the presented agent is viable and beneficial for pregnant women and healthcare providers. We intend to focus more on Natural Language Processing (NLP) techniques to improve the chatbot conversation aspects in future studies.