With the rapid development of generative artificial intelligence （AI） technology， the application of natural language processing models such as ChatGPT has become possible in the medical field. However， the issues of ethics， law， and regulation involved in medical intelligent decision-making have gradually become prominent. This paper analyzed the application and potential risks of ChatGPT in the medical field， focusing on challenges such as data privacy protection， decision fairness， and legal adaptability. It also proposed regulatory strategies， including formulating a legal framework for intelligent medical decision-making， establishing ethical norms and guidelines， strengthening regulatory mechanisms and supervisory measures， enhancing the practical ability for intelligent decision-making， and other aspects， to promote the sustainable development of medical intelligent decision-making and ensure the balance of ethics， law， and social responsibility.

Objective:To evaluate changes in operational effectiveness after the implementation of ambulatory surgical management in pars plana vitrectomy (PPV).Methods:A retrospective clinical study. 17 528 surgeries in 10 895 eyes of 10 895 patients who underwent minimally invasive PPV on an ambulatory and/or inpatient basis at Tianjin Medical University Eye Hospital from August 2015 to June 2023 were included in this study. Among them, 5 346 eyes in 5 346 cases were male; 5 549 eyes in 5 549 cases were female. The age ranged from 0 to 95 years, with the mean age of (57.74±13.15) years. 6 381 surgeries in 3 615 eyes from August 2015 to December 2018 (the initial period of day surgery) were used as the control group; 11 147 surgeries in 7 280 eyes from January 2019 to June 2023 (the expanded period of day surgery) were used as the observation group. According to the management mode of ambulatory surgery, the observation group was subdivided into the decentralized management group (January 2019 to December 2020) and the centralized management group (January 2021 to June 2023), with 2 905 and 4 375 eyes and 4 646 and 6 501 surgeries, respectively. Changes in the percentage of day surgery, average hospitalization days, and average unplanned reoperation rate were compared. The Mann-Whitney U test was used to compare numerical variables between groups; the chi-square test or Fisher's exact test was used to compare categorical variables. Results:The number of cases of daytime PPV performed in the observation group and control group was 7 852 (70.44%, 7 852/11 147) and 24 (0.38%, 24/6 381) cases, respectively, and the average hospitalization days were 1 (1) and 5 (3) d. Compared with the control group, the observation group had a significantly higher percentage of day surgery ( χ2=8 051.01) and a considerably lower mean hospitalization day ( Z=4 536 844.50), and the differences were statistically significant ( P<0.000 1). The mean hospitalization days in the decentralized and centralized management groups were 2 (3) and 1 (0) d, respectively, and unplanned reoperations were 34 (0.73%, 34/4 646) and 171 (2.63%, 171/6 501) eyes, respectively. Compared with the decentralized management group, average hospitalization days was significantly lower ( Z=1 436.94) and unplanned reoperation rate was significantly higher ( χ2=54.10) were significantly lower in the centralized management group, both of which were statistically significant ( P<0.000 1). Conclusion:PPV ambulatory management model can significantly reduce the average hospitalization day, but also results in higher rates of unplanned reoperations.

Objective:To evaluate the accuracy, completeness, and reproducibility of domestic open-source large language models (LLM) in diabetic retinopathy (DR) patient education, and to explore their potential as intelligent virtual assistants for DR patient education.Methods:A total of 41 questions and answers related to the diagnosis and treatment of DR in five categories, namely risk factors, screening and examination, symptoms and staging, diagnosis, treatment and prognosis.All questions were repeated twice as a " new dialogue" in the LLM, and all the answers were recorded.Three senior fundus physicians independently evaluated the answers on a 6-point Likert scale for accuracy and a 3-point Likert scale for completeness and repeatability, and for each answer, the evaluator was asked to make a recommendation between the LLM and the manual answers.Five questions were randomly selected to evaluate the three open source LLM, ERNIE Bot 3.5, Qwen and Kimi chat, and the LLM with the best overall performance was selected for further evaluation in the full question bank.Results:Among the three LLM, Kimi chat had the best overall performance, Kimi chat performed best, with percentages of 6 for accuracy, 3 for completeness, and 3 for repeatability among the 5 questions at 90%, 90%, and 100%, respectively.For all questions answered, the number of words in manual replies was 106 (70, 202), which was significantly lower than 505 (386, 600) in Kimi chat ( Z=-7.866, P<0.001).There was no significant correlation between the number of Kimi chat replies and the accuracy score ( rs=-0.044, P=0.492), but it was positively correlated with the integrity score ( rs=0.239, P<0.001).The interclass correlation coefficient for accuracy and completeness scores were above 0.700 among three evaluators, with the highest agreement for repeatability at 0.853, followed by completeness of the first response at 0.771.The proportion of responses ≥5 points for accuracy was 87.0%(214/246), the proportion ≥2 points for completeness was 98.0%(241/246), and the proportion higher than 70% for repeatability was 78.5%(193/246).Kimi chat excelled in answering basic questions about the disease such as disease definition, staging, frequency of screening, and common risk factors, but performed poorly on questions involving treatment choices that require a doctor's professional judgment.The proportion of evaluators choosing Kimi chat responses as superior was 69.5% (171/246), and the reasons for non-selection included lack of characteristic answers, inclusion of too much irrelevant information, and lack of responses to questions requiring a high degree of medical expertise. Conclusions:Kimi chat answers DR-related diagnostic questions in a detailed and well-organized manner, with a high degree of accuracy, completeness and reproducibility.

Objective To investigate the feasibility of developing a grading diagnostic model for schistosomiasis-induced liver fibrosis based on B-mode ultrasonographic images and clinical laboratory indicators. Methods Ultrasound images and clinical laboratory testing data were captured from schistosomiasis patients admitted to the Second People’s Hospital of Duchang County, Jiangxi Province from 2018 to 2022. Patients with grade I schistosomiasis-induced liver fibrosis were enrolled in Group 1, and patients with grade II and III schistosomiasis-induced liver fibrosis were enrolled in Group 2. The machine learning binary classification tasks were created based on patients’radiomics and clinical laboratory data from 2018 to 2021 as the training set, and patients’radiomics and clinical laboratory data in 2022 as the validation set. The features of ultrasonographic images were labeled with the ITK-SNAP software, and the features of ultrasonographic images were extracted using the Python 3.7 package and PyRadiomics toolkit. The difference in the features of ultrasonographic images was compared between groups with t test or Mann-Whitney U test, and the key imaging features were selected with the least absolute shrinkage and selection operator (LASSO) regression algorithm. Four machine learning models were created using the Scikit-learn repository, including the support vector machine (SVM), random forest (RF), linear regression (LR) and extreme gradient boosting (XGBoost). The optimal machine learning model was screened with the receiver operating characteristic curve (ROC), and features with the greatest contributions to the differentiation features of ultrasound images in machine learning models with the SHapley Additive exPlanations (SHAP) method. Results The ultrasonographic imaging data and clinical laboratory testing data from 491 schistosomiasis patients from 2019 to 2022 were included in the study, and a total of 851 radiomics features and 54 clinical laboratory indicators were captured. Following statistical tests (t = −5.98 to 4.80, U = 6 550 to 20 994, all P values < 0.05) and screening of key features with LASSO regression, 44 features or indicators were included for the subsequent modeling. The areas under ROC curve (AUCs) were 0.763 and 0.611 for the training and validation sets of the SVM model based on clinical laboratory indicators, 0.951 and 0.892 for the training and validation sets of the SVM model based on radiomics, and 0.960 and 0.913 for the training and validation sets of the multimodal SVM model. The 10 greatest contributing features or indicators in machine learning models included 2 clinical laboratory indicators and 8 radiomics features. Conclusions The multimodal machine learning models created based on ultrasound-based radiomics and clinical laboratory indicators are feasible for intelligent identification of schistosomiasis-induced liver fibrosis, and are effective to improve the classification effect of one-class data models.

Diabetes retinopathy (DR) is listed as one of the chronic diseases that should be focused on in the "14th Five-Year" National Eye Health Plan (2021-2025). Early screening is one of the effective measures to reduce blindness caused by DR. Establishing an efficient and practical community screening model is a powerful guarantee for completing early screening. The Ocular Fundus Diseases Group of the Ophthalmology Branch of the Chinese Medical Association has led the development of Expert consensus on community screening of diabetic retinopathy among DR community screening experts that is suitable for the current national situation, in order to guide and promote the further improvement of DR community screening work in China. This Expert Consensus provides detailed specifications on the current domestic trend of DR, the necessity of screening, the role of artificial intelligence grading, screening process, and quality control. This interpretation further emphasizes the importance of DR community screening, while emphasizing the responsibilities of different departments in the screening process. Finally, recommendations are provided for the sustainability of DR community screening. It is hoped that the screening rate of DR in China can be improved and blindness can be reduced by DR through Expert consensus on community screening of diabetic retinopathy and interpretation of the content.

Objective:To observe the correlation between blood cell-related inflammatory markers and diabetic retinopathy (DR).Methods:A cross-sectional study. From June 2020 to February 2022, the phase Ⅰ data of Beichen Eye Study in Tianjin Medical University Eye Hospital were included in the study. The research contents included questionnaires, routine systemic and ocular examinations, and laboratory blood cell-related indicators including mean platelet volume (MPV), platelet distribution width (PDW), neutrophils, and lymphocytes were performed. Neutrophil/lymphocyte ratio (NLR) and platelet/lymphocyte ratio (PLR) were calculated. The diagnosis and classification of DR referred to the international clinical classification standard of DR. Monocular or binocular DR was defined as DR patients. Participants were categorized into different groups based on whether they had diabetes and whether they had DR. The groups included the no-diabetes group, the diabetes without DR group, and the DR group. The Kruskal-Wallis H test was used for the comparison of quantitative data among multiple groups. Wilcoxon test was used for comparison between the two groups. The χ2 test was used to compare the categorical variables between groups. The variables was adjusted step by step, an unadjusted univariate model was built and the different parameters of the model Ⅰ, Ⅱ, Ⅲ were adjusted. The correlation between MPV, PDW, NLR, PLR, and DR in different models was analyzed by logistic regression. The receiver operating characteristic (ROC) curve was used to analyze the diagnostic efficacy of different NLR models for DR. Results:A total of 3 328 subjects were recruited. Among them, 1 121 (33.68 %, 1 121/3 328) were males and 2 207 (66.32 %, 2 207/3 328) were females. The median age of the included participants was 61.84 (6.05) years. The no-diabetes group, the diabetes without DR group, and the DR group were 2 679, 476, and 173, respectively. There was no significant difference in MPV and PLR among the three groups ( H=5.98, 1.94; P=0.051, 0.379). However, compared with no-diabetes group and the diabetes without DR group, PDW and NLR in the DR group showed an upward trend. In model Ⅲ with completely adjusted related factors, NLR was an independent risk factor for DR in no-diabetes group and DR group [odds ratio ( OR)=1.440, 95% confidence interval ( CI) 1.087-1.920, P=0.041], diabetes without DR group and DR group [ OR=1.990, 95% CI 1.440-2.749, P<0.001]. The results of ROC curve analysis showed that the diagnostic efficiency of NLR model Ⅲ was the highest, the area under the curve was 0.751 (95% CI 0.706-0.796, P<0.001), the optimal cutoff value was 0.390, and the sensitivity and specificity were 74.3% and 64.8%, respectively. Conclusions:The NLR of the DR group is significantly higher than that of the no-diabetes group and diabetes without DR group. NLR is an independent risk factor for DR.

Objective To compare the efficacy of ultrasound-guided microwave ablation versus thyroidectomy in the treatment of benign thyroid nodules (BTN) and to explore the influencing factors of prognosis. Methods A total of 282 BTN patients were enrolled and divided into ablation group (treated with ultrasound-guided microwave ablation, n=172) and resection group (treated with thyroidectomy, n=110) based on treatment methods. Perioperative indicators and thyroid function indices before and after surgery were compared between the two groups. Patients were followed up for 1 year and divided into good prognosis group [volume reduction rate (VRR)≥80%, n=210] and a poor prognosis group (VRR < 80%, n=72) based on the VRR. Factors influencing the prognosis of BTN patients were analyzed. Results The ablation group had shorter operative time and hospital stay, less intraoperative blood loss, and lower postoperative Visual Analogue Scale (VAS) scores compared to the resection group (P < 0.05). One month after surgery, serum thyrotropin (TSH) levels were higher, while free triiodothyronine (FT3) and free thyroxine (FT4) levels were lower than preoperative levels in both groups, however, TSH level was lower, and FT3 and FT4 levels were higher in the ablation group compared to the resection group (P < 0.05). Univariate analysis showed that preoperative TSH levels, the proportion of solid nodules, and the proportion of nodules in the risk zone were lower, while preoperative FT4 levels were higher in the good prognosis group compared to the poor prognosis group (P < 0.05). Binary Logistic regression analysis revealed that nodule type, nodule location, preoperative TSH, and preoperative FT4 were all influencing factors for the prognosis of BTN patients (P < 0.05). Conclusion Compared to thyroidectomy, ultrasound-guided microwave ablation for BTN offers advantages such as less intraoperative blood loss, milder postoperative pain, shorter operative time and hospital stay, and less damage to thyroid function. Preoperative TSH and FT4 levels, nodule type, and nodule location are closely related to the prognosis of BTN patients and should be given attention in clinical practice.

Objective:To compare and analyze the application of anti-vascular endothelial growth factor (VEGF) drugs for intravitreal injection in the real world before and after the establishment of one-stop intravitreal injection center, as well as the advantages and disadvantages of different management modes.Methods:A retrospective clinical study. A total of 4 015 patients (4 659 eyes) who received anti-VEGF drugs for ocular fundus diseases at the Tianjin Medical University Eye Hospital from July, 2018 to June, 2022 were included in the study. There were 2 146 males and 1 869 females. The ocular fundus diseases in this study were as follows: 1 090 eyes of 968 patients with wet age-related macular degeneration (wAMD); 855 eyes of 654 patients with diabetic macular edema (DME); 1 158 eyes of 980 patients with diabetic retinopathy (DR); 930 eyes of 916 patients with macular edema secondary to retinal vein occlusion (RVO-ME). A total of 294 eyes of 275 patients with choroidal neovascularization secondary to pathological myopia (PM-CNV); 332 eyes of 222 patients with other fundus diseases. A total of 13 796 anti-VEGF needles were injected. A total of 1 252 patients (1 403 eyes) from July 2018 to June 2020 were regarded as the control group. From July 2020 to June 2022, 2 763 patients (3 256 eyes) who received anti-VEGF treatment in the intravitreal injection center were regarded as the observation group. The total number of intravitreal injection needles, the distribution of anti-VEGF therapy in each disease according to disease classification, the proportion of patients who chose the 3+ on-demand treatment (PRN) regimen and the distribution of clinical application of different anti-VEGF drugs were compared between the control group and the observation group. The waiting time and medical experience of patients were investigated by questionnaire. χ2 test was used to compare the count data between the two groups, and t test was used to compare the measurement data. Results:Among the 13 796 anti-VEGF injections in 4 659 eyes, the total number of anti-VEGF drugs used in the control and observation groups were 4 762 and 9 034, respectively, with an average of (3.39±3.78) and (2.78±2.27) injections per eye ( t=6.900, P<0.001), respectively. In the control and observation groups, a total of 1 728 and 2 705 injections of anti-VEGF drugs were used for wAMD with an average of (5.14±4.56) and (3.59±2.45) injections per eye, respectively; a total of 982 and 2 038 injections of anti-VEGF drugs were used for DME with an average of (4.36±4.91) and (3.24±2.77) needles per eye, respectively. Additionally, a total of 942 and 2 179 injections of anti-VEGF drugs were injected for RVO-ME with an average of (3.98±3.71) and (3.14±2.15) injections per eye, respectively; a total of 291 and 615 injections of anti-VEGF drugs were injected for PM-CNV with an average of (3.31±2.63) and (2.99±1.69) injections per eye, respectively. A total of 683 and 1 029 injections of anti-VEGF drugs were injected for DR with an average of (1.60±1.26) and (1.41±1.05) injections per eye, respectively. The clinical application and implementation of "3+PRN" treatment were as follows: 223 (66.4%, 223/336) and 431 eyes (57.2%, 431/754) in the wAMD ( χ2=8.210, P=0.004), 75 (33.3%, 75/225) and 236 (37.5%, 236/630) eyes in the DME ( χ2=1.220, P>0.05), and 97 (40.9%, 97/237) and 355 eyes (51.2%, 355/693) in the RVO-ME ( χ2=7.498, P=0.006), 39 (44.3%, 39/88) and 111 eyes (53.9%, 111/206) in the PM-CNV ( χ2=2.258, P>0.05), respectively. In addition, the results of the questionnaire survey showed that there were significant differences between the control and observation groups regarding the time of appointment waiting for surgery ( t=1.340), time from admission to entering the operating room on the day of injection ( t=2.780), time from completing preoperative treatment preparation to waiting for entering the operating room ( t=8.390), and time from admission to discharge ( t=6.060) ( P<0.05). Conclusions:The establishment of a one-stop intravitreal injection mode greatly improved work efficiency and increased the number of injections. At the same time, the compliance, waiting time, and overall medical experience of patients significantly improved under centralized management.

Objective:To investigate the prevalence and risk factors of tessellation fundus (TF) among Tianjin Medical University students with different refractive statuses.Methods:A cross-sectional study. From September to December 2019, 346 students from Tianjin Medical University were randomly selected and underwent slit-lamp examination, non-cycloplegic auto-refraction, subjective refraction, best-corrected visual acuity, ocular biometric measurement, and non-dilation fundus photography. The differences in the prevalence of TF in basic characteristics and ocular biometric parameters were compared. Based on the equivalent spherical (SE), refractive status was divided into the non-myopia group (SE>-0.50 D) and the myopia group (SE≤-0.50 D). The myopia group was further divided into mild myopia group (-3.00 D26 mm group. The logistic regression was used to analyze the risk factors affecting TF. Trend tests were performed for each risk factor and TF.Results:Of the 346 subjects, 324 (93.6%, 324/346) were myopia, of whom 73 (21.1%, 73/346), 167 (48.3%, 167/346), and 84 (24.3%, 84/346) were mild myopia, moderate myopia, and high myopia, respectively; 22 (6.4%, 22/346) were non-myopia. There were 294 (85.0%, 294/346) students with TF in the macula, including 9 (40.91%, 9/22), 58 (79.45%, 58/73), 145 (86.83%, 145/167), and 82 (97.62%, 82/84) in non-myopia, low myopia, moderate myopia, and high myopia group, respectively; 52 (15.0%, 52/346) students were without TF in the macula. There were statistically significant gender differences ( χ2=4.47), SE ( t=6.29), AL ( t=-8.29), anterior chamber depth ( Z=-2.62), lens thickness ( Z=-2.23), and average corneal radius ( Z=-3.58) between students with and without TF in the macula ( P<0.05). Spherical equivalent and axial length were independent risk factors for TF and its severity ( P≤0.001). With an increasing degree of myopia, and increasing axial length, the risk of TF increased ( P for trend<0.001). Conclusions:The prevalence of TF is 85.0% among Tianjin Medical University students. TF is detected in the fundus of no myopia, mild myopia, moderate myopia and high myopia. The degree of myopia is higher, the AL is longer, the possibility of TF is higher.

Objective:To quantify any correlation between the severity of spinal curvature of an adolescent with idiopathic scoliosis and their cardiopulmonary exercise endurance.Methods:The cardiopulmonary exercise test (CPET) results and the full-length spinal X-rays in a standing position of 64 adolescents with idiopathic scoliosis were reviewed retrospectively. Independent t-tests were used to compare the two datasets obtained from those with left or right thoracic scoliosis. The correlation between the Cobb angle and cardiopulmonary exercise endurance was analyzed using Pearson correlation coefficients, multiple factor linear regression and two-stage linear regression.Results:After adjusting for gender, age, height and weight, the multiple linear regression analysis showed that the Cobb angle was significantly negatively correlated with maximum tidal volume (β=-0.013) and significantly positively correlated with the rate of respiration (β=0.421). The relationship between the Cobb angle and cardiopulmonary exercise endurance was non-linear. With a Cobb angle > 34°, a 1° increase reduces cardiopulmonary exercise endurance by a factor of 1.4 on average. At smaller Cobb angles the corresponding increase is about 0.87 times.Conclusions:The Cobb angle is a negative predictor of ventilation during exercise among adolescents with idiopathic scoliosis. The more severe a patient′s spinal curvature, the lower the cardiopulmonary exercise endurance is likely to be.