Abstract
Purpose :
In ophthalmic surgery, the most critical risks involve patient mix-up, intraocular lens (IOL) mix-up, and surgical eye mix-up. These pose significant risks of causing severe harm to patients. However, in the healthcare field, authentication has traditionally relied on human-only checks as a defense against human error. Therefore, we have conducted research and development on a mechanical check system utilizing artificial intelligence (AI). In this report, we present the results of the real-world implementation of this system (personal authentication, intraocular lens authentication, left/right eye authentication) in a clinical setting.
Methods :
The social implementation period spanned from March 4, 2022, to October 25, 2023, encompassing all cases of ophthalmic surgery performed in our department. The mechanical check system was deployed as an iPad application. Left/right eye information and IOL details were defined as correct labels during preoperative examinations by attending physicians. Personal authentication data consisted of images captured on the day of surgery before the initial examination. These correct labels were cross-verified before surgery. Monthly meetings were held to review authentication performance, and changes in empirical results were examined from the initial phase to the later stages of implementation.
Results :
Throughout the entire social implementation period, a total of 30,593 authentications for 10,501 cases were conducted. The initial implementation rates were as follows: personal authentication 99.4%, IOL authentication 88.1%, and left/right eye authentication 73.6%. In the final phase, the authentication rates were as follows: personal authentication 99.7%, IOL authentication 100.0%, and left/right eye authentication 99.8%. During the implementation period, incorrect authentications occurred in 0.248% (26 cases out of 10,501 cases).
Conclusions :
To improve the implementation rates, continuous feedback based on numerical information was essential. The mechanical check system recorded all authentication cases, contributing to the reduction of potential risks within the operating room.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.