Objective: To analyze the distribution characteristics of school bullying related contents on the Douyin platform, and to provide a scientific basis for the prevention and intervention of school bullying in the online environment. Methods: Web crawler technology was employed to collect data related to school bullying from Douyin in May 2025, using 21 keywords including "school violence" and "child being beaten". A dataset comprising video titles, publisher types, and interaction metrics was constructed. After manual screening and classification, the content themes, engagement characteristics, and sentiment tendencies of titles were analyzed. Results: A total of 2 388 valid videos released between January 2019 and May 2025 were included. Physical bullying (767 cases, 32.12%) and verbal bullying (534 cases, 22.36%) were the most common types. Content themes differed significantly across account types ( χ 2=782.85, P <0.01). Official media accounts predominantly focused on case reports ( AR = 14.52 ) and legal updates ( AR =5.49), and social media influencer accounts preferred scenario reenactments ( AR =4.07) and film clips ( AR =3.83), while ordinary users focused on coping strategies ( AR =8.76). Account types had a stronger effect on engagement metrics than content themes (all P <0.01). Sentiment analysis of titles revealed that official media accounts tended toward negative sentiment ( AR = 6.41), social media influencer accounts toward neutral sentiment ( AR =2.47), and ordinary users toward positive sentiment ( AR =4.44). Keyword analysis showed high frequency terms related to responsible parties such as "children" "parents", and "classmates" , while demand oriented words like "how" were prominent. Conclusions School bullying related contents on Douyin demonstrate significant dissemination stratification by account type, and the dissemination effect is significantly influenced by the type of content publisher. There is a clear public demand for practical guidance on bullying prevention and intervention strategies.

Objective: To analyze the occurrence of unintentional injuries among college students and their association with sexual orientation and gender identity, so as to provide a targeted scientific basis for injury prevention measures and intervention strategies. Methods: From October 24 to November 18, 2023, a sample of 1 629 college students from two general universities in Beijing was selected using convenience sampling method. A questionnaire survey was conducted to collect information on the gender identity, sexual orientation and occurrence of unintentional injuries among college students in the past year. The Symptom Checklist-90 (SCL-90), Pittsburgh Sleep Quality Index (PSQI), Childhood Trauma Questionnaire (CTQ), and Delaware Bullying Victimization Scale-Student (DBVS-S) were used to assess mental health, sleep quality, childhood trauma, and dysfunctional impulsivity status. Analyses of sexual orientation and gender identity were conducted. The t-test and Chi square test were used for intergroup comparison,and multivariate Logistic regression analysis was used to examine risk factors for unintentional injuries among college students of different gender identities. Results: The incidence rate of unintentional injuries among college students was 16.94%, with boys (17.08%) being higher than girls (16.90%). Compared with those who did not experience unintentional injuries (5.28± 3.60 , 118.68±41.38), college students who experienced unintentional injuries had poorer sleep quality and mental health status ( 6.38 ±3.93, 135.59±50.96)( t =-3.92, -4.26); the differences in the incidence of unintentional injury among college students with non suicidal self injury, interpersonal violence, childhood trauma, and different sexual orientations and gender identities were all statistically significant ( χ 2=28.75, 75.18, 9.83, 16.20, 4.13) (all P <0.05). Multivariate Logistic regression analysis showed that after adjusting for age, gender and body mass index, non heterosexual orientation increased the risk of unintentional injuries ( OR=1.61, 95%CI =1.09-2.38), whereas existing non suicidal self injury behaviors ( OR=2.10, 95%CI =1.02-4.37) and poorer mental health status ( OR=1.54, 95%CI =1.05-2.27) increased the risk of unintentional injuries among non heterosexual college students (all P <0.05). Conclusions The incidence rate of unintentional injuries among college students is relatively high, with non heterosexual groups having increased risk of unintentional injuries. Mental health status and non suicidal self injury behaviors are important factors related to unintentional injuries among non heterosexual college students.

Objective:To summarize the progress in the research of textual quality evaluation of health-related media reports.Methods:A systematic literature retrieval about textual quality evaluation of health-related media reports based on PubMed, Web of Science, Embase, Wanfang database, and China National Knowledge Infrastructure was conducted. Information regarding the textual quality definition, evaluation dimensions, indicators and methods of included papers was extracted.Results:A total of 29 study papers were included in this analysis, in which 26 were about retrospective textual quality evaluation of health-related media reports, and 3 were about the model or tool development for textual quality evaluation of health-related media reports. The topics of news reports included: 16 studies on injury, 3 on general health, 3 on infectious disease, 3 on cancer screening and treatment, 3 on chronic non-communicable disease, and 1 on medication risk. The definition of textual quality of health-related media reports and the dimensions of the quality evaluation varied across the studies. The quality evaluation indicators of media reports can be divided into three categories: availability of surveillance information, availability of professional information, and adherence to principles of media reporting. Most studies conducted the quality evaluation manually, with only 2 studies employing semi-automated or automated evaluation methods.Conclusions:No unified definition, set of dimensions, indicators, or automated algorithms exist for evaluating the textual quality of health-related media reports, which limits assessing massive news data effectively. It is necessary to conduct methodological studies on the textual quality evaluation of health-related media reports based on journalism and communication theory, infodemiology, deep learning, natural language processing, text mining, as well as specific disease and injury prevention theory.

Objective:To summarize the progress in the research of textual quality evaluation of health-related media reports.Methods:A systematic literature retrieval about textual quality evaluation of health-related media reports based on PubMed, Web of Science, Embase, Wanfang database, and China National Knowledge Infrastructure was conducted. Information regarding the textual quality definition, evaluation dimensions, indicators and methods of included papers was extracted.Results:A total of 29 study papers were included in this analysis, in which 26 were about retrospective textual quality evaluation of health-related media reports, and 3 were about the model or tool development for textual quality evaluation of health-related media reports. The topics of news reports included: 16 studies on injury, 3 on general health, 3 on infectious disease, 3 on cancer screening and treatment, 3 on chronic non-communicable disease, and 1 on medication risk. The definition of textual quality of health-related media reports and the dimensions of the quality evaluation varied across the studies. The quality evaluation indicators of media reports can be divided into three categories: availability of surveillance information, availability of professional information, and adherence to principles of media reporting. Most studies conducted the quality evaluation manually, with only 2 studies employing semi-automated or automated evaluation methods.Conclusions:No unified definition, set of dimensions, indicators, or automated algorithms exist for evaluating the textual quality of health-related media reports, which limits assessing massive news data effectively. It is necessary to conduct methodological studies on the textual quality evaluation of health-related media reports based on journalism and communication theory, infodemiology, deep learning, natural language processing, text mining, as well as specific disease and injury prevention theory.

Objective:To analysis the influenced factors of readmission in hospital among patients with mood disorder,construct predictive models and compare the predictive performance of the models.Methods:The electron-ic medical record data of patients with mood disorder admitted to Beijing Anding Hospital from January 2010 to De-cember 2018 were retrospectively collected.Utilizing stepwise logistic regression to analyze the related factors of re-admission in patients with mood disorder.Logistic regression,support vector machine,random forest,extreme gradi-ent boosting(XGBoost)algorithms and convolutional neural networks(CNN)were used to construct a readmission prediction model and compare the predictive performances of the different models by using the accuracy,precision,recall,et.al.Results:A total of 6 234 patients with mood disorder were enrolled,24.9％(n=1 549)patients were readmitted after discharge.The stepwise logistic regression results revealed that readmissions were more likely to occur in patients with mood disorder who were female,had comorbidities,had a treatment outcome of cure,were treated with MECT and were on second-generation antipsychotics(OR=1.26,1.68,1.26,1.35,1.18).The CNN model demonstrated the highest accuracy,precision,and F1 scores,all at 0.87,with a recall of 0.86.The random forest achieved a recall of 0.86 and an AUC of 0.95.Conclusion:Readmission of patients with mood disorder is influenced by a multitude of factors.The convolutional neural networks and random forest models outperform other models in prediction.

Objective:To analysis the influenced factors of readmission in hospital among patients with mood disorder,construct predictive models and compare the predictive performance of the models.Methods:The electron-ic medical record data of patients with mood disorder admitted to Beijing Anding Hospital from January 2010 to De-cember 2018 were retrospectively collected.Utilizing stepwise logistic regression to analyze the related factors of re-admission in patients with mood disorder.Logistic regression,support vector machine,random forest,extreme gradi-ent boosting(XGBoost)algorithms and convolutional neural networks(CNN)were used to construct a readmission prediction model and compare the predictive performances of the different models by using the accuracy,precision,recall,et.al.Results:A total of 6 234 patients with mood disorder were enrolled,24.9％(n=1 549)patients were readmitted after discharge.The stepwise logistic regression results revealed that readmissions were more likely to occur in patients with mood disorder who were female,had comorbidities,had a treatment outcome of cure,were treated with MECT and were on second-generation antipsychotics(OR=1.26,1.68,1.26,1.35,1.18).The CNN model demonstrated the highest accuracy,precision,and F1 scores,all at 0.87,with a recall of 0.86.The random forest achieved a recall of 0.86 and an AUC of 0.95.Conclusion:Readmission of patients with mood disorder is influenced by a multitude of factors.The convolutional neural networks and random forest models outperform other models in prediction.

OBJECTIVES: During the epidemic of coronavirus disease-2019 (COVID-19), the wide spread of rumors caused significant public hazards. This study aims to understand the situation of discrimination for typical COVID-19 rumors by the public and related factors. METHODS: An anonymous online survey was carried out using Questionnaire Star. The contents included participants' gender, age, education level, the COVID-19 information sources, and the judgmental questions about 14 representative COVID-19 rumors. The discrimination rate and 95% confidence interval of 14 rumors were estimated, and the association of discrimination rate with gender, age, and education level was analyzed by binary logistic regression. RESULTS: A total of 2 087 valid questionnaires were collected. The participants were mainly female (62.7%) and below 35 years old (63.4%); the education level was predominantly college/bachelor's degree (47.3%) and master's degree or above (39.1%); the participants, who accessed to COVID-19 information included internet media, accounted for 91%. The participants with different gender, age, and education level had significant differences in the distribution of COVID-19 information sources (all P<0.01). The participants' discrimination rate for 14 rumors ranged from 67.4% to 98.6%, with 4 rumors less than 80%. Women's discrimination rate of 9 rumors was significantly higher than men's (all P<0.05). There was no significant difference in the discrimination rate of rumors among the different age groups (all P>0.05), but the differences in the discrimination rate of other rumors among the different age groups varied according to the rumor. Compared to those with high school or less education levels, the discrimination rates were also higher in the respondents with high education levels (P<0.05). CONCLUSIONS A few publics are still unable to identify typical rumors during the COVID-19 epidemic. There are associations among genders, age, and the education levels with the discrimination of some rumors. The government authorities should strengthen the true information regarding COVID-19, and therefore enhance the public's ability to identify rumors.
Humans ; Female ; Male ; Adult ; COVID-19/epidemiology* ; Epidemics ; Surveys and Questionnaires

Objective:To search and summarize the relevant evidences for perioperative hydration in prevention of contrast-induced acute kidney injury (CIAKI) in patients undergoing coronary intervention, and to provide evidence-based basis for clinical nursing staff to develop personalized perioperative hydration plan and prevent CIAKI.Methods:Evidence-based nursing was used to search evidences for perioperative hydration in prevention of CIAKI in patients undergoing coronary intervention in UpToDate, National Guideline Clearing-house (NGC) , European Society of Cardiology (ESC) , JBI Evidence Summary Database, Cochrane Database, PubMed, Wanfang Database, China National Knowledge Infrastructure (CNKI) and China Biomedical Literature Database (CBM) , including evidence summary, guidelines, expert consensus and systematic reviews. The search time was from the establishment of the database to February 2019. Two researchers evaluated the quality of included literatures and extracted evidence from literatures that met the quality standards.Results:A total of 9 articles of evidence were included, including 1 evidence summary, 3 guidelines, 3 expert consensus and 2 systematic evaluations. Finally, a total of 10 pieces of evidence from 5 aspects, including CIAKI risk assessment, hydration fluid type, hydration path and timing, hydration rate and urine volume monitoring, were summarized.Conclusions:This study summarizes the best evidence for perioperative hydration in prevention of CIAKI in patients undergoing coronary intervention and provides an evidence-based basis for clinical medical staff.

With the development of society and the transition of travel ways,traumatic brain injury (TBI) caused by external causes such as traffic crashes,falls,sports,recreation,etc.has become more and more common.TBI has become a serious public health problem worldwide.Understanding the epidemiology,causes,and trends of TBI is critical for prevention and rehabilitation.Unfortunately,most of the available epidemiological studies on TBI were conducted among outpatients and/or inpatients at local hospitals,and national surveys with high-quality or surveillance data are scarce in China.Thus,there is no comprehensive understanding of TBI.In order to provide the basis for future investigation and improve the effects of TBI prevention and rehabilitation,we reviewed literature on TBI published in recent 10 years and discussed the development of epidemiological studies on TBI in definition,classification,external cause,epidemiology,and health-related impacts.

[Objective] To explore the clinical effects of three methods in the treatment of cesarean scar pregnancy.[Method]All the patients with cesarean scar pregnancy accepted treatment at 2 affiliated hospitals of Sun Yat-Sen University from May 2006 to September 2016 were retrospectively analyzed.According to their different therapies,186 of them were divided into three groups:surgery group (undergone transvaginal removal of ectopic pregnancy tissue and repair of uterine defect,73 cases),medical treatment group (treated with medical and combined with curettage,51 cases),UAE group (treated with UAE and combined with suction curettage,62 cases).The success rate,hospitalization time,hospitalization expenses and treatment side effects of the three groups were compared.[Result] The success rates of the three groups were 98.6％,54.9％,and 80.6％;The hospitalization days of the three groups were (7.3 ± 3.3) days,(16.2 ± 11.2) days,(12.6 ± 10.4) d.The hospitalization expenses of the three groups were (7 967 ±1 987) yuan,(5 499-± 1 965) yuan,(15 863 ± 5 941) yuan.The treatment side effects of the three groups were 15.1％,23.5％,and 37.1％.The success rate,hospitalization days,hospitalization expenses and treatment side effects of the three groups were statistically different (P ＜ 0.05).The surgery group has a highest success rate,shortest hospitalization time,lower hospitalization expenses,and less complication.[Conclusions] Transvaginal removal of ectopic pregnancy tissue and repair of uterine defect is a reliable treatment for CSP.