Feasibility of utilizing artificial intelligence to assist junior anesthesia residents in making preoperative anesthesia plans
10.3760/cma.j.cn131073.20231018.00416
- VernacularTitle:人工智能技术辅助麻醉科低年资住院医师制订术前麻醉方案的可行性
- Author:
Lin LI
1
;
Ju GAO
;
Yali GE
;
Tingting ZHANG
;
Keshi YAN
Author Information
1. 扬州大学临床医学院 江苏省苏北人民医院麻醉科 225000
- Keywords:
Artificial intelligence;
Anesthesia department, hospital;
Internship and residency;
Anesthesia;
Preoperative period
- From:
Chinese Journal of Anesthesiology
2024;44(4):461-465
- CountryChina
- Language:Chinese
-
Abstract:
Objective:To evaluate the feasibility of utilizing artificial intelligence (AI) to assist junior anesthesia residents in making the preoperative anesthesia plans.Methods:Forty anesthesia residents in their third year of training, who had obtained their practicing physician qualifications in the Yangzhou area, were assigned into 4 groups ( n=10 each) using a random number table method: Chat-GPT combined with Bing chat group (C-G-B group), Chat-GPT group (C-G group), Bing chat group (B group), and control group (C group). Fifty patients undergoing elective non-cardiac surgery were selected from the anesthesia clinic as teaching cases. C-G-B, C-G and B groups utilized different AI tools to assist trainees in designing anesthesia plans, producing standardized textual outputs. Each trainee underwent a baseline knowledge test through a professional theory examination prior to enrollment. The completeness and accuracy of the preoperative anesthesia plans were evaluated and scored by 3 chief anesthesiologists. The total time spent on plan formulation and satisfaction scores regarding AI tool feedback were recorded. An analysis was conducted based on the American Society of Anesthesiologists (ASA) Physical Status classification of teaching cases. Results:In ASA Physical Status classification Ⅰ and Ⅱ teaching cases, there was no statistically significant difference in completeness and accuracy scores among the four groups ( P>0.05). In ASA Physical Status classification Ⅲ teaching cases, compared to C group, the completeness and accuracy scores were significantly increased in C-G-B, C-G and B groups, with the highest scores observed in C-G-B group ( P<0.05). Among all teaching cases (ASA Physical Status classification Ⅰ-Ⅲ), the total time spent was significantly shortened in C-G and B groups as compared to C and C-G-B groups ( P<0.05). There was no statistically significant difference in the total time spent between C-G group and C-G-B group ( P>0.05). Compared to C-G and B groups, the satisfaction score was significantly decreased in C-G-B group ( P<0.05). Conclusions:For ASA Physical Status classification Ⅲ patients, using AI to assist junior anesthesia residents in making preoperative anesthesia plans may offer advantages. Although combining the use of Chat-GPT and Bing chat can further improve the completeness and accuracy of anesthesia plan development, it may require more time.
