DOI: 10.4103/ijb.ijb_8_25 ISSN: 0971-653X

How accurately do artificial intelligence-based voice assistants address questions about burn injuries? An analytical study

Hetal Amipara, Akesh Gopalakrishnan, Shubhashree Page, Shubham Rajesh Pathode, Shyam Popat

Purpose:

Artificial intelligence (AI) is a rapidly advancing field with significant applications in healthcare, including voice-activated virtual assistants (VAs) such as Amazon’s Alexa, Google Assistant, Microsoft’s Copilot, and Apple’s Siri. These VAs have potential utility in providing information and support to patients, including those with burn injuries. This study evaluates the effectiveness of four widely used VAs in answering commonly asked questions regarding burns.

Materials and Methods:

Four VAs were assessed: Siri (VA1), Google Assistant (VA2), Alexa (VA3), and Copilot (VA4). These VAs were queried using ten frequently asked questions about burn injuries. Responses were rated using a modified Likert scale, where a score of 1 indicated a high-quality, adequate voice response and a score of 5 showed the lowest. One-way analysis of variance and Tukey’s post hoc analyses were conducted to compare the Vas’ performance.

Results:

Significant differences were found in the average scores per question among the VAs. VA4 (1.2 ± 0.15) achieved the best performance, followed by VA2 (1.62 ± 0.45), VA1 (3.67 ± 0.35), and VA3 (3.82 ± 0.65). The combined mean scores revealed a highly significant difference ( P < 0.001) among the VAs, except between VA4 and VA2 ( P = 0.123) and VA1 and VA3 ( P = 0.580).

Conclusion:

VA4 proved to be the most effective in answering burn injury-related questions, followed by VA2, VA1, and VA3. The integration of advanced AI in VAs has the potential to enhance patient care, but further studies are needed to address real-world variability and expand the scope of inquiries.

More from our Archive