Blood transfusion is an important life-support measure in modern medicine. As clinical demand for transfusion continues to rise, blood supply remains under persistent strain, making the efficient utilization of this precious medical resource a critical component of blood management. Accurate prediction of transfusion needs is of great significance for optimizing blood resource allocation and conserving blood products. Traditional transfusion decision-making relies on clinical experience and simplified scoring systems, which are insufficient to meet the demands of precision medicine. Machine learning (ML) technology can integrate multidimensional data—including demographic characteristics, laboratory indicators, surgical information, and dynamic physiological waveforms—to construct high-performance predictive models, offering a new approach to transfusion prediction. This article systematically reviews the application progress of ML in transfusion prediction across clinical scenarios such as trauma, the perioperative period, and obstetrics. In trauma, ML has been applied to early massive transfusion prediction, prehospital transfusion decision-making, and precise prediction in pediatric trauma. In the perioperative field, applications span specialties including traumatic brain injury, orthopedic surgery, cardiovascular and major vascular surgery, and hepatic surgery. In obstetrics, models can effectively predict postpartum hemorrhage and transfusion requirements associated with cesarean section. Studies have shown that algorithms such as random forest, gradient boosting machines, and deep neural networks achieve area under the receiver operating characteristic curve (AUC) values of 0.83-0.98 in massive transfusion prediction, 0.75-0.97 in perioperative scenarios, and 0.80-0.89 in obstetric transfusion prediction—significantly outperforming traditional scoring tools. However, current studies are generally limited by single-center retrospective designs, insufficient external validation, and inadequate reporting standards; limited model interpretability and challenges in clinical integration further constrain practical translation. Future research should focus on constructing multicenter data cohorts, integrating multimodal data, conducting prospective implementation studies, and deeply embedding models within electronic health record systems, thereby advancing the application of ML to optimize clinical transfusion decision-making.

Objective:To explore the efficacy and its related factors of rituximab (RTX) in the treatment of children with frequently relapsing nephrotic syndrome/steroid-dependent nephrotic syndrome (FRNS/SDNS).Methods:It was a single-center retrospective study. The clinical data of FRNS/SDNS children first treated with RTX in the First Affiliated Hospital of Sun Yat-sen University from November 1, 2016 to September 1, 2023 were collected. The number of relapse within 1 year before and after RTX treatment, the time to first relapse after RTX treatment, and the time to B-cell reconstitution were analyzed. At the first treatment, a single dose of RTX was given at 375 mg/m 2, with a maximum dose of 500 mg, once a week, for 1 to 4 doses. The count of CD19 + lymphocytes in the peripheral blood of the children was continuously monitored. If B-cell reconstruction was performed, the decision on whether to proceed to the next course of RTX treatment was made based on clinical manifestations. Kaplan-Meier method was used to analyze relapse-free survival rate after receiving RTX. Cox proportional hazards regression model was used to analyze the related factors of relapse after RTX treatment. Results:A total of 98 FRNS/SDNS children receiving RTX treatment were enrolled, including 75 males (76.5%). The age at onset was 4.0 (1.9, 7.1) years and age of receiving RTX was 11.3 (8.5, 13.5) years. There were 90 children (91.8%) achieving complete remission, while 8 patients (8.2%) did not respond to RTX treatment, and 3 patients (3.1%) progressed to end-stage kidney disease after receiving RTX. The relapse-free survival rates at 6 months and 1 year after RTX treatment were 83.3% (75/90) and 57.9% (22/38), respectively. The frequency of relapse 1 year after RTX treatment decreased compared to 1 year before RTX treatment ( Z=-7.398, P<0.001). Compared with children without relapse during the period of B-cell depletion, relapsed children had a higher number of relapse within one year after RTX treatment ( Z=5.246, P<0.001). The time to first relapse after RTX treatment was 8.3 (4.6, 13.9) months in 51 relapse patients. Compared with children receiving 1 dose of RTX in the first course, those receiving 2 or more doses had a longer time to the first relapse ( Z=2.983, P=0.003). There was no statistically significant difference in time to the first relapse between children who received mycophenolate mofetil therapy after RTX treatment and those who didn't ( P>0.05). The reconstruction time of B cells after the first course of RTX was 6.9 (5.3, 9.0) months. Compared to children receiving one dose of RTX in the first course, those receiving two or more doses had a longer B-cell reconstitution time ( Z=2.739, P=0.006). There was no statistically significant difference in B-cell reconstitution time between children who received mycophenolate mofetil therapy after RTX treatment and those who didn't ( P>0.05). Univariate Cox regression analysis showed that recurrence after calcineurin inhibitor (CNI) treatment before RTX treatment and the number of recurrence in one year before RTX treatment were correlated factors of recurrence after RTX treatment (both P<0.05). Multivariate Cox regression analysis showed that recurrence after CNI treatment before RTX treatment was an independent correlated factor of relapse after RTX therapy ( HR=3.496, 95% CI 1.245-9.818, P=0.018). Infusion reactions occurred in 10 patients (10.2%) and infections were observed in 24 patients (24.5%) during B cell depletion. No serious adverse events occurred. Conclusions:RTX is well tolerated and effective in treating FRNS/SDNS. Recurrence after CNI treatment before RTX treatment may be an independent related factor of relapse after RTX treatment.

Objective:To analyze the association between remnant cholesterol (RC) and the risk of atherosclerotic cardiovascular disease (ASCVD) in community population in Shanghai.Methods:Using baseline and follow-up data from the Shanghai Suburban Adult Cohort and Biobank, individuals with ASCVD (including coronary heart disease, stroke, myocardial infarction, and peripheral artery disease) at baseline were excluded. A Cox proportional hazards regression model was employed to analyze the relationship between RC and ASCVD risk and the association under different LDL-C levels.Results:A total of 57 281 participants were included, with a median follow-up of 5.61 person-years. During the follow-up, 1 436 ASCVD events (2.51%) were recorded. After adjusting for potential confounders, individuals with moderate ( HR=1.18, 95% CI: 1.03-1.36) or high RC levels ( HR=1.32, 95% CI: 1.15-1.51) had an increased risk of ASCVD. The association was stronger in participants younger than 60 years-old (interaction P=0.048). Participants with RC ≥0.97 mmol/L and LDL-C <3.40 mmol/L demonstrated a 19% ( HR=1.19, 95% CI: 1.06-1.35) increased risk of ASCVD. When RC ≥0.97 mmol/L and LDL-C ≥3.40 mmol/L, ASCVD risk increased by 42% ( HR=1.42, 95% CI: 1.21-1.67). Conclusions:Elevated RC increases ASCVD risk, regardless of LDL-C levels. RC can serve as a valuable predictor and intervention target for ASCVD.

Objective:To analyze the mediating effect of cardiovascular health status (CVH) on the association between educational level and cardiovascular disease (CVD).Methods:The participants were from Shanghai Suburban Adult Cohort and Biobank, and questionnaire survey, physical examination, blood biochemistry were conducted from 2016 to 2020 for baseline information collection, and follow up was conducted until March 31, 2024 based on the medical data, CVD incidence data and death surveillance data at different levels. The associations of educational level, CVH and time to CVD onset of the study population were analyzed using the accelerated failure time model to analyze the mediating effects of CVH, health behaviors, and health factors in the association of educational level and time to CVD onset. The mediating effects of educational level, gender, and age moderated associations were also analyzed.Results:A total of 57 312 participants were included, with 2 780 new cases of CVD during a median follow-up of 6.71 (6.71-6.72) years, and a mean incidence density of 7.77/1 000 person-years (95% CI: 7.48/1 000 person-years -8.06/1 000 person-years). In total, the less educational level and the lower CVH, the higher CVD incidence density ( P<0.05). The results of accelerated failure time models showed that the time ratio for CVD-free survival was 1.15 (95% CI: 1.06-1.24) and 1.33 (95% CI: 1.10-1.60) for moderate and high educational level, respectively. The results of the mediation effect analysis showed that the association between moderate and high educational level and time to CVD onset was 29.60% (20.50%-50.00%) and 36.10% (23.80%-59.00%), 9.97% (5.07%-20.00%) and 13.84% (6.84%-29.00%), 15.24% (9.64%-27.00%) and 17.55% (11.58%-33.00%) of mediators mediated by CVH, health behaviors, health factors, respectively. Among them, there was an exposure-mediated interaction of educational level and a positive moderating effect of age. Conclusion:CVH, health behaviors and health factors had a proportionate mediating effect in the association between educational level and risk of CVD development.

Objective To construct a classification model for surgical patients and apply it in the allocation of nursing human resources,providing a reference for nursing human resource management.Methods A convenience sampling method was used to retrospectively select 5,431 hospitalized surgical patients admitted to 6 surgical nursing units of a tertiary general hospital in Suzhou from July to November 2022 as the subjects of this study.The nursing hours were measured,and related influencing factors were analyzed.A decision tree classification method was used to establish a classification model for surgical patients.From August to October 2022,1,527 hospitalized surgical patients admitted to 3 nursing units of the same hospital were conveniently selected.The minimum number of nurses required daily was calculated using the surgical patient classification model,actual nursing hours measurement method,nurse-to-bed ratio method,and 8-hour continuous shift scheduling method.The application effect of the surgical patient classification model in nursing human resource allocation was evaluated with the actual nursing hours measurement method as the standard.Results The surgical patient classification model includes 7 classification indicators:length of hospital stay,diagnosis-related group weight,presence or absence of secondary care orders,surgical grade,anesthesia method,age,and presence or absence of critical illness orders.Patients were divided into 14 groups,and the model explained 90.5％of the total variance in nursing workload.The minimum number of nurses required in surgical nursing units calculated based on this model was closest to the result of the actual nursing hours measurement method and was superior to the results of the nurse-to-bed ratio method and the 8-hour continuous shift scheduling method.Conclusion The surgical patient classification model can accurately reflect the nursing workload of such patients.The classification indicators are simple and easy to obtain,and can guide the allocation of human resources in surgical nursing units.

Objective:To observe the effects of shoulder-three-needle therapy plus joint mobilization and extracorporeal shock wave therapy(ESWT)in improving symptoms and pain mediators in patients with scapulohumeral periarthritis.Methods:A total of 82 patients with scapulohumeral periarthritis were randomized into a control group and an observation group by the random number table method,with 41 cases in each group.The control group was treated with ESWT.The observation group was treated with additional shoulder-three-needle therapy plus joint mobilization based on the treatment in the control group.After 4 weeks of treatment,shoulder joint function,pain mediators,inflammatory factor levels,and clinical efficacy were compared between the two groups.Results:The total effective rate and the score of shoulder joint function in the observation group were higher than those in the control group(P<0.05).After treatment,the levels of serum pain mediators substance P(SP),5-hydroxytryptamine(5-HT),and prostaglandin E2(PGE2),as well as inflammatory factors interleukin(IL)-6,tumor necrosis factor(TNF)-α,and C-reactive protein(CRP)decreased significantly in both groups(P<0.05).And,the serum pain mediators SP,5-HT,and PGE2,and the levels of inflammatory factors IL-6,TNF-α,and CRP in the observation group were all lower than those in the control group(P<0.05).Conclusion:Shoulder-three-needle therapy plus joint mobilization and ESWT has a better curative effect than ESWT alone in treating scapulohumeral periarthritis.It can effectively improve the clinical symptoms,reduce the levels of pain mediators and inflammatory factors,and increase shoulder joint function.

Objective:To construct a biosafety evaluation index system for major emergency public health events in medical institutions.Methods:Based on previous laboratory biosafety evaluation work, relevant regulations and standards on biosafety in China were collected through literature research and expert consultations. Candidate indicators for constructing the biosafety evaluation system for major emergency public health events in medical institutions were selected, and a framework was established. Two rounds of expert questionnaires were conducted to determine the content of the index system based on experts′ evaluation, and each indicator′s relevance and importance were scored. Finally, two rounds of Delphi consultations were carried out, and the Analytic Hierarchy Process (AHP) was applied to calculate the weights of indicators.Results:The response rates for the total four rounds of questionnaire surveys were all 100%. The first two rounds focused on determining the framework, while the latter two focused on determining the weights for each indicator. The authority coefficients of the expert consultations for the two rounds of weights were 0.65 and 0.70, respectively, indicating the reliability of the research results. In the final round of survey, the Kendall′s coefficients of concordance at each level were all greater than 0.1. Through statistical testing, the P-values were all less than 0.05, indicating good coordination of expert opinions. Ultimately, we established an operational biosafety evaluation system for major emergency public health events in medical institutions, consisting of 4 primary indicators, 26 secondary indicators, and 119 tertiary indicators, with additional deduction items, bonus items, unacceptable items, and monitoring indicators.Conclusions:Based on scientific theory, a biosafety evaluation system for major emergency public health events in medical institutions was constructed, achieving the integration of routine and emergency measures. This system can be used for self-assessment of laboratory biosafety during emergency public health events, addressing the lack of unified standards in biosafety evaluation. Through regular self-assessment, it can enhance the level of biosafety management in medical institution laboratories, to realize the value of application and dissemination.

Objective To analyze the characteristics of facial and tongue chromaticity parameters in patients with gastrointestinal tumors by setting the inspection image characteristics of patients with gastrointestinal tumors as the main research content;To establish an auxiliary diagnostic model for gastrointestinal tumors.Methods One-way ANOVA,t-test,Mann-whitney U test,canonical correlation analysis and Spearman statistical methods were used to analyze the characteristics of inspection image indexes and correlation of tumor markers of the 391 cases in the control group and 359 patients with gastrointestinal tumors.Machine learning methods such as SVM,Random Forest,KNN,Naive Bayes,XG Boost and Ada Boost were used to establish an auxiliary diagnostic model for gastrointestinal tumors.Results In terms of facial indicators,there were differences in F-R,F-G and F-B indicators among the control group,early-stage gastrointestinal cancer patients,and mid-to late-stage gastrointestinal cancer patients,in the comparison of tongue features among,TC-L,TB-L and TB-a of the control group,patients with early gastrointestinal tumors,and patients with intermediate and advanced gastrointestinal tumors showed a gradual downward trend;the AUC of the auxiliary diagnosis model of gastrointestinal tumor disease based on the chromaticity parameters of face tongue image constructed by Ada Boost algorithm was 0.930.Conclusion The auxiliary diagnostic model of gastrointestinal diseases constructed by facial and tongue images has good diagnostic effect,which can provide objective data support for in-depth exploration of the complex relationship between diagnosis and disease.

IgA nephropathy（IgAN）is the most common primary glomerular disease，characterized by the deposition of IgA in the glomerular mesangial area. IgAN is a progressive disease with a protracted and relapsing course. Approximately 50%~75% of adult IgAN patients experience renal function deterioration and eventually progress to end-stage kidney disease（ESKD）within 20 years after diagnosis. Among patients with onset in childhood，about 30% ultimately progress to ESKD after 20 years of long-term follow-up. In recent years，numerous studies have continuously explored important signaling pathways and key molecules in its pathogenesis. This review introduces the therapeutic advances in IgAN，aiming to provide better treatment options for clinicians.

Objective : To explore the role of lysophosphatidylcholine acyltransferase 3(LPCAT3) in Erastin-induced ferroptosis of acute myeloid leukemia(AML) cells and its related molecular regulatory mechanisms. Methods : Tetrazolium salt(MTS) method was used to detect the sensitivity of different AML cells to the classic ferroptosis inducer Erastin, real time quantitative polymerase chain reaction(qPCR) was used to detect the basal expression level ofLPCAT3mRNA, and the correlation between them was analyzed. Lentivirus-mediatedLPCAT3overexpression AML cell lines(OE group) and negative control lines(NC group) were constructed. After Erastin intervention, MTS, flow cytometry, and micromethods were used to detect cell viability, lipid reactive oxygen species(ROS), and Malondialdehyde(MDA), respectively. qPCR and Western blot were used to detect unfolded protein response(UPR) classic pathway signaling molecules(PERK, ATF4, GRP78, etc.) expression levels. The above ferroptosis-related indicators were detected after combined intervention with the UPR inhibitor 4-phenylbutyric acid(4-PBA), and the regulatory relationship was analyzed. Results : Four different types of AML cells had different sensitivities to ferroptosis, among which K562 cells were relatively insensitive. The IC50of the four types of AML cells to Erastin was negatively correlated with the expression level ofLPCAT3(r=-0.919,P<0.001). After Erastin intervention, the cell viability of K562 cells in the OE group was significantly inhibited by Erastin compared with the NC group(P<0.001), and the levels of lipid ROS and MDA increased(P<0.001). The results of qPCR and Western blot showed that, compared with the NC group, the mRNA and protein expression of UPR classic pathway moleculesPERK,ATF4, andGRP78mRNA and protein increased in the OE group(P<0.01). After inhibiting the UPR pathway by 4-PBA, the viability of K562 cells decreased(P<0.01), and lipid ROS and MDA levels increased(P<0.01) compared with the uninhibited state. Conclusion Overexpression ofLPCAT3can promote ferroptosis in K562 cells, and this process is negatively regulated by the classical UPR pathway PERK/ATF.