Development And Application Of Acceptance Scales For AI In Endoscopy By Endoscopists And Subjects

Background: In recent years,digestive endoscopy has developed rapidly in China,and new technologies such as endoscopic early cancer screening and minimally invasive treatment have emerged in endlessly.However,the mastering level of new endoscopic diagnosis and treatment technologies in various regions and hospitals is uneven.The emergence of artificial intelligence(AI)assistant system in endoscopy(hereinafter referred to as endoscopic AI)is expected to solve this problem.In recent years,more and more endoscopic AI has been developed at home and abroad,and clinical application and effect evaluation have been carried out.However,there is no research on the acceptance of endoscopic AI and its influencing factors by endoscopists and subjects.How to evaluate the acceptance of endoscopic AI by endoscopists and subjects? There is a lack of quantitative measurement tools at home and abroad.This study developed two acceptance scales for AI in endoscopy by endoscopists and subjects for the first time,to measure the acceptance of endoscopic AI by endoscopists and subjects,and then the scales have been preliminarily applied to explore the influencing factors of endoscopic AI acceptance by endoscopists and subjects,which may provide important reference for the research,development,application and promotion of endoscopic AI.Methods:(1)Scale development: The primary scales for measuring the acceptance of endoscopic AI by endoscopists and subjects were developed by literature survey,research group discussion and semi-structured questionnaire investigation.The primary scales were modified based on Delphi method and then we obtained the first edition of the scales.The scales were used to conduct electronic and paper questionnaire investigations on endoscopists who have used endoscopic AI in three endoscopy centers in China and endoscopic subjects in the Third Xiangya Hospital of Central South University.SPSS 25.0 software was used to test the reliability and validity of the scales based on data from questionnaire investigations.The reliability is evaluated by Cronbach’s α and half coefficient(Spearman-Brown).The content validity is carried out through expert consultation,the calibration validity is evaluated by the correlation between the scale score and the calibration,and the structural validity is evaluated by exploratory factor analysis.According to the results of reliability and validity test and respondents’ feedback,the scales were modified through research group discussion.Finally,the scales for endoscopic AI acceptance by endoscopists and subjects were developed.(2)Scale verification and application: The Acceptance Scale for Endoscopic AI by Endoscopists was used to conduct questionnaire investigation among the endoscopists who had used endoscopic AI in 10 endoscopic centers in China.The Acceptance Scale for Endoscopic AI by Endoscopic Subjects was used to conduct questionnaire investigation among endoscopic subjects in the Third Xiangya Hospital of Central South University.The reliability and validity of the scales were evaluated by Cronbach’s α and confirmatory factor analysis by AMOS 24.0.The total scores and dimension scores of the scales were calculated based on questionnaire investigations to explore the acceptance of endoscopic AI by endoscopists and subjects.The chi square test and binary Logistic regression analysis were used to explore the influencing factors on acceptance of endoscopic AI by endoscopists and subjects,and further explore the impact of these factors on the scores of each dimension.Results:(1)The development and application of the Acceptance Scale for Endoscopic AI by Endoscopists:(1)Scale development: the scale includes4 dimensions(medical service,psychological emotion,dependence,accuracy),15 items.210 valid questionnaires were recovered.The total Cronbach’s α was 0.904;the correlation coefficient between the total score of the scale and the calibration was 0.889;in the construct validity,the KMO(Kaiser Meyer Olkin)value was 0.893.The dimensions and items of the scale are set reasonably with good reliability and validity,indicating that the scale can accurately measure the acceptance of endoscopic AI by endoscopists.(2)Scale verification and application: The scale was used to investigate endoscopists who had used endoscopic AI in 10 endoscopic centers in China,and 322 valid questionnaires were collected.The total Cronbach’s α was 0.917;in the structural validity,the CMIN/DF was 3.392,and the root mean square error of approximation(RMSEA)was 0.086.The results show that 68.9% endoscopists had a high acceptance of endoscopic AI,and the scores of four dimensions(medical service,psychological emotion,dependence and accuracy)of the scale were 3.44,3.71,3.68 and3.85 respectively.Analyzing the influencing factors of the acceptance of endoscopic AI by endoscopists,the results showed that: Endoscopists with lower professional title(OR=2.027,95%CI:1.316～3.122,P<0.001)and more cases of using endoscopic AI(OR=2.381,95%CI:1.430～3.963,P=0.001)have a higher acceptance of endoscopic AI.However,gender,hospital level had no significant effect on the acceptance of endoscopic AI by endoscopists.Further analysis of the factors affecting the scores of each dimension shows that: The lower scores of dimensions 1(medical service),2(psychological emotion)and 3(dependence)are related to the higher professional title and the fewer cases of endoscopic AI use,indicating that endoscopists with the higher professional title and the fewer cases of endoscopic AI use have the lower awareness of the positive role of endoscopic AI in medical service and psychological emotion,and they are more worried about dependency on AI.The lower scores of dimensions 4(accuracy)is related to the lower professional title and the shorter duration of using endoscopic AI,indicating that endoscopists with the higher professional title and the shorter duration of using endoscopic AI have the lower awareness of the endoscopic AI’s accuracy.The value of endoscopic AI for medical service is the key point of popularizing endoscopic AI.At the same time,it is necessary to further strengthen endoscopists’ awareness of endoscopic AI and conducting endoscopic AI training.(2)The development and application of the Acceptance Scale for Endoscopic AI by Endoscopic Subjects:(1)Scale development: the scale includes 3 dimensions(accuracy,ethics,benefit and willingness),11 items.351 valid questionnaires were recovered.The total Cronbach’s α was 0.864;the correlation coefficient between the total score of the scale and the calibration was 0.636;in the construct validity,the KMO(Kaiser Meyer Olkin)value was 0.788.The dimensions and items of the scale are set reasonably with good reliability and validity,indicating that the scale can accurately measure the acceptance of endoscopic AI by endoscopic subjects.(2)Scale verification and application: The scale was used to investigate endoscopic subjects in the Third Xiangya Hospital of Central South University and 335 valid questionnaires were collected.The total Cronbach’s α was 0.848;in the structural validity,the CMIN/DF was 3.774,and RMSEA was 0.091.The results show that 52.2% endoscopic subjects had a high acceptance of endoscopic AI and the scores of the three dimensions(accuracy,ethics,benefit and willingness)of the scale were3.53,3.60 and 3.71 respectively.Analyzing the influencing factors of the acceptance of endoscopic AI by endoscopic subjects,the results showed that: Endoscopic subjects with higher health literacy(OR=1.779,95%CI:1.140-2.776,P=0.011)have a higher acceptance of endoscopic AI,while gender,age,education experience,industry,work status,annual family income,and awareness of AI had no significant independent impact on their acceptance of endoscopic AI.Further analysis of the factors affecting the scores of each dimension showed that: the lower score of dimension 1(accuracy)was related to the higher education,which indicated that the subjects who have received higher education have lower recognition of the accuracy of endoscopic AI.The lower score of dimension 2(ethics)is related to the older age and the lower health literacy,indicating that subjects with the older age and the lower health literacy are more worried about ethical issues related to endoscopic AI.The lower score of dimension 3(benefit and willingness)was related to the lower awareness of AI,indicating that subjects with lower awareness of AI have lower awareness of the benefits of endoscopic AI.To promote endoscopic AI,it is necessary to improve and inform the subjects of the ethical system related to endoscopic AI to reduce their concerns;At the same time,strengthen publicity to increase the subjects’ awareness of endoscopic AI.Conclusion:(1)This study successfully developed two scales to measure the acceptance of endoscopy AI by endoscopists and subjects for the first time,which provides reliable tools for the promotion of endoscopic AI.(2)Endoscopists with lower professional title and more cases of using endoscopic AI have a higher acceptance of endoscopic AI.Subjects with high health literacy have a high acceptance of endoscopic AI.Inadequate awareness of endoscopic AI is the main reason that restricts endoscopists and subjects to accept endoscopic AI.This study provided guidance for the promotion of endoscopic AI in the future.