Feasibility of a Virtual Assistant for Patients' Frequently Asked Questions: An Unexplored Artificial Intelligence Application in Plastic Surgery
Daniel Boczar1, David J Restrepo1, Andrea Sisti1, Maria T. Huayllani1, Jeremie D. Oliver2, Haytham Helmi1, Colleen Ball1, Brian D RInker1, Antonio Jorge Forte1.
1Mayo Clinic Florida, Jacksonville, FL, USA, 2Mayo Clinic School of Medicine, Rochester, MN, USA.
Virtual Assistants (VA) is a segment of Artificial Intelligence (AI) that is rapidly developing. However, its utilization to address patients' frequently asked questions (FAQs) preoperatively remains unexplored. We hypothesize that a VA could address preoperative FAQs related to plastic surgery procedures. We developed a VA and assessed its accuracy and participants' opinion regarding the answers and the technology.
Using IBM Watson Assistant Platform, we developed a VA to answer 10 topics of plastic surgery patients' FAQs. Between July and August of 2018, we recruited subjects with administrative positions at our health care institution to chat with the VA. They asked, with their own words, 1 question for each topic and filled out a satisfaction questionnaire. Post-survey analysis of questions and answers allowed assessment of the VA's accuracy. Data collected was described in frequencies and percentages.
30 participants completed the survey. The majority was female (70%), and the mean age was 27.76 years old (SD 8.68, 19 to 51years old). The overall accuracy of the plastic surgery virtual assistant was 92.3% (277/294 questions), and participants considered the answer correct in 83.3% of the times (250/294 answers). Most of the participants considered the VA easy to use, answered adequately, and could be helpful for patients. However, when asked if it could replace a human assistant, they stayed neutral.
AI is predicted to save $300-450 billion annually in the American Health Care System, but the feasibility of VA to address patients' FAQs was never been assessed before. This pilot study demonstrated that it is feasible, and that volunteers reported that the VA could be helpful for patients.
Back to 2019 Abstracts