With the rapid advancement and iterative development of new artificial intelligence technologies, there remains a regulatory vacuum in corresponding governance measures among governments worldwide. Simultaneously, a technological and governance gap exists between developing countries and developed economies. In response, the World Health Organization (WHO) has released "Ethics and Governance of Artificial Intelligence for Health: Guidance on Large Multi-Modal Models" to assist governments in strengthening governance capabilities in this field. This paper provides an in-depth analysis of the Guidance, aiming to identify challenges and risks associated with the application of multimodal large models in healthcare. Guided by ethical principles for advancing health through artificial intelligence, the paper examines the three-tier governance framework and recommendations outlined in the Guidance. Additionally, it evaluates the current state of AI governance in China, offering insights and reference points for improving AI governance in China′s healthcare sector.

Objective:To evaluate the diagnostic performance of artificial intelligence-assisted diagnostic systems in cervical cytopathological examination.Methods:Cervical cytology slide data were retrospectively collected from four hospitals for the external validation of the developed artificial intelligence-assisted diagnostic system. Subsequently, prospective data collection was conducted for human-machine assisted studies.Results:In the retrospective study, a total of 3 162 valid samples were collected as external validation data. The system showed an area under the curve (AUC) of 0.890 (95% CI: 0.878-0.902), accuracy of 0.885 (95% CI: 0.873-0.896), sensitivity of 0.928 (95% CI: 0.914-0.941), and specificity of 0.852 (95% CI: 0.834-0.867). In the prospective study, 212 valid samples were collected, and five junior cytologists participated in the human-machine assisted study. Without artificial intelligence assistance, the average AUC for the five cytologists was 0.686 (95% CI: 0.650-0.722), the accuracy was 0.699 (95% CI: 0.671-0.727), the sensitivity was 0.653 (95% CI: 0.599-0.703), the specificity was 0.719 (95% CI: 0.685-0.750), the Fleiss κ value was 0.510, and the reading time was 223 seconds. With artificial intelligence assistance, the AUC, accuracy, sensitivity, and specificity increased by 0.166, 0.143, 0.225, and 0.107, respectively. Additionally, Fleiss κ was 0.730 and the reading time decreased by 188 seconds. All differences were statistically significant (all P<0.001). Conclusions:Artificial intelligence-assisted diagnosis system shows excellent performance and good generalizability, significantly improving the diagnostic accuracy, consistency, and efficiency of junior cytologists. It can be an effective auxiliary tool for junior cytologists in clinical practice.

With the rapid advancement and iterative development of new artificial intelligence technologies, there remains a regulatory vacuum in corresponding governance measures among governments worldwide. Simultaneously, a technological and governance gap exists between developing countries and developed economies. In response, the World Health Organization (WHO) has released "Ethics and Governance of Artificial Intelligence for Health: Guidance on Large Multi-Modal Models" to assist governments in strengthening governance capabilities in this field. This paper provides an in-depth analysis of the Guidance, aiming to identify challenges and risks associated with the application of multimodal large models in healthcare. Guided by ethical principles for advancing health through artificial intelligence, the paper examines the three-tier governance framework and recommendations outlined in the Guidance. Additionally, it evaluates the current state of AI governance in China, offering insights and reference points for improving AI governance in China′s healthcare sector.

Objective:To evaluate the diagnostic performance of artificial intelligence-assisted diagnostic systems in cervical cytopathological examination.Methods:Cervical cytology slide data were retrospectively collected from four hospitals for the external validation of the developed artificial intelligence-assisted diagnostic system. Subsequently, prospective data collection was conducted for human-machine assisted studies.Results:In the retrospective study, a total of 3 162 valid samples were collected as external validation data. The system showed an area under the curve (AUC) of 0.890 (95% CI: 0.878-0.902), accuracy of 0.885 (95% CI: 0.873-0.896), sensitivity of 0.928 (95% CI: 0.914-0.941), and specificity of 0.852 (95% CI: 0.834-0.867). In the prospective study, 212 valid samples were collected, and five junior cytologists participated in the human-machine assisted study. Without artificial intelligence assistance, the average AUC for the five cytologists was 0.686 (95% CI: 0.650-0.722), the accuracy was 0.699 (95% CI: 0.671-0.727), the sensitivity was 0.653 (95% CI: 0.599-0.703), the specificity was 0.719 (95% CI: 0.685-0.750), the Fleiss κ value was 0.510, and the reading time was 223 seconds. With artificial intelligence assistance, the AUC, accuracy, sensitivity, and specificity increased by 0.166, 0.143, 0.225, and 0.107, respectively. Additionally, Fleiss κ was 0.730 and the reading time decreased by 188 seconds. All differences were statistically significant (all P<0.001). Conclusions:Artificial intelligence-assisted diagnosis system shows excellent performance and good generalizability, significantly improving the diagnostic accuracy, consistency, and efficiency of junior cytologists. It can be an effective auxiliary tool for junior cytologists in clinical practice.

The rapid development of artificial intelligence in the field of healthcare has greatly improved diagnosis accuracy, disease prediction, personalized treatment and healthcare resource management. However, with the widespread application of medical artificial intelligence, challenges has emerged in the aspects of medical data, model development and evaluation, and societal considerations. Therefore, this study aims to explore challenges in the application of artificial intelligence in healthcare and suggest a series of feasible solutions to improve medical professional and researchers' understanding of medical artificial intelligence and enhance the quality of healthcare in clinical practice

Objective: To explore the most economically feasible cervical cancer screening strategies in urban China. Methods: A series of Markov models were constructed to evaluate health and economic outcomes of different screening strategies. There were 24 screening strategies including four screening methods: liquid-based cytology (LBC), human papillomavirus (HPV) DNA genotyping, HPV DNA genotyping with LBC triage (HPV DNA+ LBC), HPV DNA genotyping and LBC co-testing (HPV DNA-LBC), along with three intervals (every 1, 3 or 5 years) and two starting age for screening (30 or 35 years old) were compared. Models parameters were obtained from a cervical cancer screening study in urban China and literature reviews. Results: The cumulative incidence and mortality risk of cervical cancer declined over 69% and 82% respectively for each screening strategy as compared with the no screening scenario. LBC every five years starting from 35 years old strategy cost the least (RMB 690 per capita) and could save life years compared with no screening. The cost effectiveness ratios of 24 strategies ranged from -10 903 to 117 992 RMB per life year saved. All strategies were cost-effective compared to no screening. In the incremental cost-effectiveness analysis, LBC every 5 years starting from 30 strategy, HPV DNA genotyping every 3 years starting from 30 strategy, LBC every 3 years starting from 30 strategy and LBC every year starting from 30 strategy were dominant strategies. Conclusions Screening can effectively prevent cervical cancer. In urban Chinese areas with insufficient socioeconomic resources, LBC every 5 years from 35 years old strategy is recommended. In relatively more affluent areas, LBC every 5 years from 30 years old strategy, LBC every 3 years from 30 years old strategy, HPV DNA genotyping every 3 years from 30 years old strategy, and LBC every year from 30 years old strategy are recommended successively.

Objective: To study the mRNA expressions of various CD97 isoforms in colorectal carcinoma tissues and their clinical significances. Methods: A total of 50 colon cancer patients in the First Affiliated Hospital of Wenzhou Medical University from December 2013 to May 2014 and human colon cancer cell lines SW480 and SW620 were enrolled. The real-time fluorescence quantitative polymerase chain reaction (qRT-PCR) was used to detect the mRNA expressions of CD97 human epidermal growth factor (EGF) (1, 2, 5), CD97EGF (1, 2, 3, 5) and CD97EGF (1, 2, 3, 4, 5) in colon cancer tissues, adjacent tissues, normal colon tissues, SW480 cells and SW620 cells. The relationship between the mRNA expression of CD97EGF (1, 2, 5) and the clinicopathological factors was analyzed. Results: Compared with those low expressions in adjacent tissues and normal tissues, the mRNA expressions of CD97 isoforms CD97EGF (1, 2, 5), CD97EGF (1, 2, 3, 5) and CD97EGF (1, 2, 3, 4, 5) in cancer tissues were highest, and the differences were statistically significant (0.71±0.20 vs. 0.40±0.09 vs. 0.35±0.07, F = 107.642, P < 0.01; 0.45±0.11 vs. 0.26±0.05 vs. 0.27±0.06, F = 94.231, P < 0.01; 0.41±0.10 vs. 0.21±0.05 vs. 0.19±0.03, F = 165.672, P < 0.01). In addition, the mRNA expression of CD97EGF (1, 2, 5) in colon cancer patients was associated with tumor infiltration depth (T₁-T₂ and T₃-T₄), clinical stages (Ⅰ-Ⅱ and Ⅲ-Ⅳ), and the differences were statistically significant (t = -2.582, P = 0.013; t = -5.062, P < 0.01). The mRNA expression of CD97EGF (1, 2, 5) in SW620 cells was higher than that in SW480 cells. Conclusions CD97 isoforms are highly expressed in colon cancer tissues, and CD97EGF (1, 2, 5) may play an important role in the development and invasion of colon cancer. The CD97 isoforms may be new markers in the treatment of colon cancer.

Objective: To evaluate 3D imaging technology in the preoperative evaluation of breast conserving surgery. Methods: A ret-rospective analysis was conducted using clinical data from 38 patients who underwent breast conserving surgery that was assisted by 3D imaging technology in Xiaogan Hospital Affiliated to Wuhan University of Science and Technology from April 2017 to January 2019. All 38 patients underwent 3.0-T breast magnetic resonance imaging (MRI) examination before surgery, and 3D reconstruction of virtu-al images was constructed through 3D modeling of medical digital imaging and communication (DICOM) data. The predicted resected tissue volume was compared with the volume of the actual resected specimen, and the surgical margin and postoperative aesthetics of the breast conserving surgery were evaluated. Results: The reconstructed 3D model clearly displayed the anatomical structures of the breast, tumor, gland, and blood vessels, and their relationship in 3D spaces. The goodness of fit of the 3D model to the practical sit-uation was 97.4% (37/38). In terms of the resection tissue volume, there was no significant difference between the predicted results (PRTV) and actual results (ARTV) [(61.7 ± 20.1) mL vs. (65.1 ± 20.7) mL, P>0.05]. There was a strong positive correlation between ARTV and PRTV (P<0.01). One patient underwent supplementary secondary surgery, resulting in an incidence of 2.6% (1/38). The postopera-tive satisfaction for breast conserving surgery was 100% (38/38). Conclusions: 3D imaging technology clearly displays the anatomical relationship between breast tumor and surrounding tissues, and correctly assesses breast volume, guiding surgical resection.

Objective: To account the direct cost of uterine cervix carcinoma treatment in China and to explore the related factors which influence the direct financial burden of the disease. Methods: Data was collected through the medical record system and telephone interviews in 14 county-level hospitals and 9 provincial and municipal hospitals from 14 provinces/municipalities enrolled in the Chinese National Health Industry Research Project in 2015. The direct financial burden of uterine cervix carcinoma treatment consisted of the direct medical cost and the direct non-medical cost of treatment in different pathological cervical cancer stages and precancerous lesions. Multiple liner regression method was used to analyze the factors affecting the costs. Results: The age of the 3 246 patients was (46.40±10.43) years, including 2 423 patients from provincial and municipal hospitals and 823 patients from county-level hospitals. The direct financial burden for one patient of pathological uterine cervix carcinoma stage or precancerous lesion ranged from 10 156.3 yuan to 75 716.4 yuan in provincial and municipal hospitals, and for patients from county-level hospitals, the cost was between 4 927.9 yuan and 47 524.8 yuan per person. There was a wide gap between the direct financial burden of patients in different disease stages. The direct financial burden of patients with precancerous lesions ranged from 4 927.9 yuan per person to 11 243.0 yuan per person, as for patients of pathological uterine cervix carcinoma stages, the direct financial burden was between 29 274.6 yuan and 75 716.4 yuan per person. The factors which influence direct financial burden would include: the levels of the hospital, pathological period, medicare reimbursement, days of treatment, and the methods of treatment (P<0.001). Conclusion The direct financial burden of diseases in patients with pathological uterine cervix carcinoma stage or precancerous lesion differed in different levels of hospital and pathological periods. In addition, medicare reimbursement, days of treatment, and the methods of treatment all had impact on it.

Through a brief overview of the origin and development of artificial intelligence, the research progress of artificial intelligence in digestive endoscopy, ophthalmoscopy and electronic colposcopy was summarized, and the importance of its application was clarified. This article induced the main problems and suggestions of artificial intelligence in the development of medical endoscopes, and prospected the future application of artificial intelligence in medical endoscopy diagnosis.