ABSTRACT

Since the inception of ChatGPT in 2022, it has demonstrated promise in a number of medical applications, including supporting laboratory and clinical diagnosis, remote patient monitoring, and giving healthcare practitioners information about new developments and updates. Furthermore, it holds promise in increasing the rate of digital medicine amongst patients, as well as offering prompt assistance in clinic practice. This study will evaluate the accuracy of this language model in providing information on myopia,and its capacity to provide patients and medical professionals with verified data on myopia. The study employed the use of a cross-sectional questionnaire survey that was conducted in Nigeria. 398 practicing Optometrists participated and evaluated the answers generated by ChatGPT to frequently asked questions on myopia. A closed-ended Likert scale questionnaire was used for this study and distributed through paper-based, and soft copy format using Google form. The findings gotten were analyzed using the Statistical Program for Social Sciences (SPSS), version 22.0 (IBM SPSS Inc., Chicago, IL, USA). Overall, majority of the responses were considered good (40.41%), 35.49% were considered acceptable, and 14.53% of the responses were considered to be very good. Only a small proportion of the responses were considered to be very poor (2.26%) and 7.3% were rated poor by the participants. Considerable inaccuracies were seen in the symptoms, where headache was rated to be poor (42%), and in treatment where opticians were rated poor in managing myopia (37.9%). Using Pearson’s Chi-square method, a statistical significance (p<0.05) was observed between the different socio-demographics and rating of the responses provided; gender (p=0.000), place of practice (p=0.000), duration of practice (p=0.027), state of practice (p=0.000), and primary source of Optometry-related information (0.000). Although this cross-sectional study showed promising outcomes in using ChatGPT to answer common myopia-related patient inquiries, more research on their possible impact in clinical situations is necessary before any firm conclusions about ChatGPT can be drawn.  

Keywords: Artificial intelligence, Chat GPT, Large Language Model, Natural Language Processing, myopia.