Objective: To investigate the factors influencing intraoperative blood transfusion in patients with hip fractures and to develop a machine learning (ML) model for predicting this risk. Methods: A total of 424 patients with hip fractures who underwent surgical treatment between November 2022 and March 2025 in our hospital were selected. Key feature variables of intraoperative blood transfusion risk were identified using the Boruta algorithm. Four different ML algorithms—support vector machine (SVM), linear discriminant analysis (LDA), mixed discriminant analysis (MDA), and extreme gradient boosting (XGBoost)—were used to develop predictive models for intraoperative blood transfusion risk. The predictive performance of the four ML models were evaluated using accuracy, precision, receiver operating characteristic (ROC) curves, precision-recall curves (PRC), precision-recall gain curves (PRGC), and F1 scores. Shapley additive interpretation (SHAP) was used to interpret the final model. Results: Among the 424 patients, 77(18.2%) received intraoperative blood transfusion. The Boruta algorithm identified albumin (ALB), activated partial thromboplastin time (APTT), types of anesthesia, types of fracture, and hemoglobin (Hb) as key feature variables for predicting intraoperative blood transfusion risk. In model evaluation, the SVM model outperforms the other three models across multiple metrics, including the area under the receiver operating characteristic curve (AUC), recall, recall gain, accuracy, precision, F1 score, and the area under the precision-recall curve (PRC-AUC). The SVM model, interpreted and visualized based on SHAP values, effectively predicted intraoperative blood transfusion risk in patients with hip fracture. A visual online application was developed based on the SVM model (https://pbo-nomogram.shinyapps.io/blood/). Conclusion: Preoperative low ALB and Hb levels, prolonged APTT, general anesthesia, and intertrochanteric fractures are risk factors for intraoperative blood transfusion in hip fracture patients. The risk prediction model for intraoperative blood transfusion constructed based on the SVM algorithm has optimal performance, which provides new ideas and methods for the clinical early identification of hip fracture patients with high transfusion risk and the implementation of targeted interventions.

Background Lung cancer, one of the most common malignant tumors worldwide, has long ranked first in cancer incidence and mortality, posing a severe challenge to public health systems. Objective To analyze the trends in incidence, mortality, and disability-adjusted life years (DALYs) of lung cancer among residents in Jing'an District, Shanghai, from 2002 to 2021, explore the impacts of population aging, population growth, and age-specific prevalence on disease burden, and provide a scientific basis for optimizing regional lung cancer prevention and control strategies. Methods Based on the cancer registration and cause-of-death surveillance data of registered residents in Jing'an District, Shanghai, from 2002 to 2021, Joinpoint regression models were used to analyze the annual change trends (APC) and average annual change trends (AAPC) of lung cancer incidence, mortality, DALY rate, and their age-standardized rates. Decomposition analysis was applied to quantify the contribution of population aging, population growth, and age-specific prevalence to changes in the number of new cases, deaths, and DALYs. Results From 2002 to 2021, the crude incidence rate of lung cancer in Jing'an District increased from 68.00 per 100000 to 144.36 per 100000 (AAPC=4.86%, P<0.001), and the age-standardized incidence rate rose from 43.09 per 100000 to 61.46 per 100000 (AAPC=2.89%, P<0.001). The growth rate of incidence was significantly higher in females (AAPC=6.73%) than in males (AAPC=3.62%). The crude mortality rate increased from 58.08 per 100000 to 66.11 per 100000 (AAPC=1.10%, P<0.001), while the age-standardized mortality rate decreased (AAPC=−1.54%, P<0.001). The crude DALY rate showed an upward trend (AAPC=0.94%, P=0.002), whereas the age-standardized DALY rate declined (AAPC=−1.63%, P<0.001). Significant differences were observed across genders and age groups: the mortality and DALY rates increased in males aged 60-64 years, while the fastest incidence growth occurred in females aged 45-64 years. The results of decomposition analysis indicated that the increase in new cases was primarily attributed to age-specific incidence (38.15%) and aging (37.31%); the growth in deaths and DALYs was driven by aging, while age-specific mortality and population growth showed offsetting effects. From 2002 to 2021, the probability of premature death from lung cancer among males (2.04%-2.52%) in Jing'an District showed no statistically significant trend (AAPC=−0.28%, P=0.440), while it presented a decreasing trend in the total population (1.37%-1.89%) and females (0.72%-1.26%) (AAPC=−0.80% and −2.23%, respectively; P=0.029 and <0.001, respectively). Conclusion From 2002 to 2021, the risk of lung cancer incidence in Jing’an District, Shanghai, continue to increase, with a disease burden exhibiting significant heterogeneity by gender and age: key features include a rapid increase in new cases and a trend toward younger onset among females, alongside rising mortality risk among males aged 60–64 years. Population aging and increasing age-specific incidence rates are primary drivers of the growing number of cases, while declining age-specific mortality rates mitigate the rise in mortality burden.

ObjectiveTo investigate the mechanism of action of Kaixinsan in the treatment of Alzheimer's disease （AD） based on network pharmacology， molecular docking， and animal experimental validation. MethodsThe Traditional Chinese Medicine Systems Pharmacology Database and Analysis Platform（TCMSP） and the Encyclopedia of Traditional Chinese Medicine（ETCM） databases were used to obtain the active ingredients and targets of Kaixinsan. GeneCards， Online Mendelian Inheritance in Man（OMIM）， TTD， PharmGKB， and DrugBank databases were used to obtain the relevant targets of AD. The intersection （common targets） of the active ingredient targets of Kaixinsan and the relevant targets of AD was taken， and the network interaction analysis of the common targets was carried out in the STRING database to construct a protein-protein interaction（PPI） network. The CytoNCA plugin within Cytoscape was used to screen out the core targets， and the Metascape platform was used to perform gene ontology（GO） functional enrichment analysis and Kyoto encyclopedia of genes and genomes（KEGG） pathway enrichment analysis. The “drug-active ingredient-target” interaction network was constructed with the help of Cytoscape 3.8.2， and AutoDock Vina was used for molecular docking. Scopolamine （SCOP） was utilized for modeling and injected intraperitoneally once daily. Thirty-two male C57/BL6 mice were randomly divided into blank control （CON） group （0.9% NaCl， n=8）， model （SCOP） group （3 mg·kg^-1·d^-1， n=8）， positive control group （3 mg·kg^-1·d^-1 of SCOP+3 mg·kg^-1·d^-1 of Donepezil， n=8）， and Kaixinsan group （3 mg·kg^-1·d^-1 of SCOP+6.5 g·kg^-1·d^-1 of Kaixinsan， n=8）. Mice in each group were administered with 0.9% NaCl， Kaixinsan， or Donepezil by gavage twice a day for 14 days. Morris water maze experiment was used to observe the learning memory ability of mice. Hematoxylin-eosin （HE） staining method was used to observe the pathological changes in the CA1 area of the mouse hippocampus. Enzyme linked immunosorbent assay（ELISA） was used to determine the serum acetylcholine （ACh） and acetylcholinesterase （AChE） contents of mice. Western blot method was used to detect the protein expression levels of signal transducer and activator of transcription 3（STAT3） and nuclear transcription factor（NF）-κB p65 in the hippocampus of mice. ResultsA total of 73 active ingredients of Kaixinsan were obtained， and 578 potential targets （common targets） of Kaixinsan for the treatment of AD were screened out. Key active ingredients included kaempferol， gijugliflozin， etc.. Potential core targets were STAT3， NF-κB p65， et al. GO functional enrichment analysis obtained 3 124 biological functions， 254 cellular building blocks， and 461 molecular functions. KEGG pathway enrichment obtained 248 pathways， mainly involving cancer-related pathways， TRP pathway， cyclic adenosine monophosphate（cAMP） pathway， and NF-κB pathway. Molecular docking showed that the binding of the key active ingredients to the target targets was more stable. Morris water maze experiment indicated that Kaixinsan could improve the learning memory ability of SCOP-induced mice. HE staining and ELISA results showed that Kaixinsan had an ameliorating effect on central nerve injury in mice. Western blot test indicated that Kaixinsan had a down-regulating effect on the levels of NF-κB p65 phosphorylation and STAT3 phosphorylation in the hippocampal tissue of mice in the SCOP model. ConclusionKaixinsan can improve the cognitive impairment function in SCOP model mice and may reduce hippocampal neuronal damage and thus play a therapeutic role in the treatment of AD by regulating NF-κB p65， STAT3， and other targets involved in the NF-κB signaling pathway.

ObjectiveTo establish a model that can quickly identify the aroma components in different parts of Nardostachys jatamansi， so as to provide a quality control basis for the market circulation and clinical use of N. jatamansi. MethodsHeadspace solid-phase microextraction-gas chromatography-mass spectrometry（HS-SPME-GC-MS） combined with Smart aroma database and National Institute of Standards and Technology（NIST） database were used to characterize the aroma components in different parts of N. jatamansi， and the aroma components were quantified according to relative response factor（RRF） and three internal standards， and the markers of aroma differences in different parts of N. jatamansi were identified by orthogonal partial least squares-discriminant analysis（OPLS-DA） and cluster thermal analysis based on variable importance in the projection（VIP） value >1 and P<0.01. The odor data of different parts of N. jatamansi were collected by Heracles Ⅱ Neo ultra-fast gas phase electronic nose， and the correlation between compound types of aroma components collected by the ultra-fast gas phase electronic nose and the detection results of HS-SPME-GC-MS was investigated by drawing odor fingerprints and odor response radargrams. Chromatographic peak information with distinguishing ability≥0.700 and peak area≥200 was selected as sensor data， and the rapid identification model of different parts of N. jatamansi was established by principal component analysis（PCA）， discriminant factor alysis（DFA）， soft independent modeling of class analogies（SIMCA） and statistical quality control analysis（SQCA）. ResultsThe HS-SPME-GC-MS results showed that there were 28 common components in the underground and aboveground parts of N. jatamansi， of which 22 could be quantified and 12 significantly different components were screened out. Among these 12 components， the contents of five components（ethyl isovalerate， 2-pentylfuran， benzyl alcohol， nonanal and glacial acetic acid，） in the aboveground part of N. jatamansi were significantly higher than those in the underground part（P<0.01）， the contents of β-ionone， patchouli alcohol， α-caryophyllene， linalyl butyrate， valencene， 1，8-cineole and p-cymene in the underground part of N. jatamansi were significantly higher than those in the aboveground part（P<0.01）. Heracles Ⅱ Neo electronic nose results showed that the PCA discrimination index of the underground and aboveground parts of N. jatamansi was 82， and the contribution rates of the principal component factors were 99.94% and 99.89% when 2 and 3 principal components were extracted， respectively. The contribution rate of the discriminant factor 1 of the DFA model constructed on the basis of PCA was 100%， the validation score of the SIMCA model for discrimination of the two parts was 99， and SQCA could clearly distinguish different parts of N. jatamansi. ConclusionHS-SPME-GC-MS can clarify the differential markers of underground and aboveground parts of N. jatamansi. The four analytical models provided by Heracles Ⅱ Neo electronic nose（PCA， DFA， SIMCA and SQCA） can realize the rapid identification of different parts of N. jatamansi. Combining the two results， it is speculated that terpenes and carboxylic acids may be the main factors contributing to the difference in aroma between the underground and aboveground parts of N. jatamansi.

Objective To develop an innovative recognition algorithm that aids physicians in the identification of pulmonary nodules. Methods Patients with pulmonary nodules who underwent thoracoscopic surgery at the Department of Thoracic Surgery, Affiliated Drum Tower Hospital of Nanjing University Medical School in December 2023, were enrolled in the study. Chest surface exploration data were collected at a rate of 60 frames per second and a resolution of 1 920×1 080. Frame images were saved at regular intervals for subsequent block processing. An algorithm database for lung nodule recognition was developed using the collected data. Results A total of 16 patients were enrolled, including 9 males and 7 females, with an average age of (54.9±14.9) years. In the optimized multi-topology convolutional network model, the test results demonstrated an accuracy rate of 94.39% for recognition tasks. Furthermore, the integration of micro-variation amplification technology into the convolutional network model enhanced the accuracy of lung nodule identification to 96.90%. A comprehensive evaluation of the performance of these two models yielded an overall recognition accuracy of 95.59%. Based on these findings, we conclude that the proposed network model is well-suited for the task of lung nodule recognition, with the convolutional network incorporating micro-variation amplification technology exhibiting superior accuracy. Conclusion Compared to traditional methods, our proposed technique significantly enhances the accuracy of lung nodule identification and localization, aiding surgeons in locating lung nodules during thoracoscopic surgery.

AIM: To compare the accuracy of intraocular lens(IOL)calculation formulas based on different corneal refractive power in calculating IOL diopters of cataract patients with a history of corneal refractive surgery.METHODS: A prospective clinical study was conducted with a cohort of 32 cataract patients(42 eyes)who had previously undergone myopic laser corneal surgery at Jinan Mingshui Eye Hospital between February 2022 and August 2024. The study employed several IOL calculation formulas, including the Haigis-L formula, the Barrett True K formula based on simulated keratometry(SimK), the Haigis formula based on total keratometry(TK), the Potvin-Hill Pentacam(PVP)formula based on corneal true net power(TNP), and the OCT formula based on net corneal power(NCP). These formulas were used to calculate IOL power and predict postoperative refractive outcomes. At 1 mo postoperatively, subjective refraction was performed, and the prediction error(PE), mean absolute prediction error(MAE), median absolute prediction error(MedAE), and the percentage of prediction errors within the ranges of ±0.25, ±0.50, ±0.75, and ±1.0 D were determined.RESULTS: The intraclass correlation coefficient for the four types of corneal refractive power was 0.986(P<0.001). There was no significant difference between TNP and NCP(P=0.491), and there were differences between the other two groups(all P<0.001). Statistically significant differences were observed between PE and 0 for the Haigis-L(K)and Haigis(TK)formulas(all P<0.001). In contrast, no statistically significant differences were noted between PE and 0 for the PVP, OCT, and Barrett True K formulas(all P>0.05). The MedAE value of Barrett True K was the smallest 0.32(0.19, 0.71)D among the five formulas, and there was no significant difference in MedAE among the five formulas(P=0.870). The proportion of eyes with PE within ±0.25 and ±1.0 D in Barrett True K formula was 38%(16/42)and 95%(40/42), respectively. The proportion of eyes within ±0.50 D in PVP formula was 71%(30/42); and the proportion of eyes with PE within ±0.75 D in Haigis(TK)formula was 83%(35/42).CONCLUSION: After corneal refractive surgery, there are differences between different types of corneal refractive power. When calculating IOL, the accuracy of TK combined with Haigis formula is better than that of Haigis-L(K)formula, and Barrett True K formula shows good accuracy.

Purpose: The genomic characteristics of uterine sarcomas have not been fully elucidated. This study aimed to explore the genomic landscape of the uterine sarcomas (USs). Materials and Methods: Comprehensive genomic analysis through RNA-sequencing was conducted. Gene fusion, differentially expressed genes (DEGs), signaling pathway enrichment, immune cell infiltration, and prognosis were analyzed. A deep learning model was constructed to predict the survival of US patients. Results: A total of 71 US samples were examined, including 47 endometrial stromal sarcomas (ESS), 18 uterine leiomyosarcomas (uLMS), three adenosarcomas, two carcinosarcomas, and one uterine tumor resembling an ovarian sex-cord tumor. ESS (including high-grade ESS [HGESS] and low-grade ESS [LGESS]) and uLMS showed distinct gene fusion signatures; a novel gene fusion site, MRPS18A–PDC-AS1 could be a potential diagnostic marker for the pathology differential diagnosis of uLMS and ESS; 797 and 477 uterine sarcoma DEGs (uDEGs) were identified in the ESS vs. uLMS and HGESS vs. LGESS groups, respectively. The uDEGs were enriched in multiple pathways. Fifteen genes including LAMB4 were confirmed with prognostic value in USs; immune infiltration analysis revealed the prognositic value of myeloid dendritic cells, plasmacytoid dendritic cells, natural killer cells, macrophage M1, monocytes and hematopoietic stem cells in USs; the deep learning model named Max-Mean Non-Local multi-instance learning (MMN-MIL) showed satisfactory performance in predicting the survival of US patients, with the area under the receiver operating curve curve reached 0.909 and accuracy achieved 0.804. Conclusion USs harbored distinct gene fusion characteristics and gene expression features between HGESS, LGESS, and uLMS. The MMN-MIL model could effectively predict the survival of US patients.

BackgroundModified electroconvulsive therapy （MECT） is a common front-line strategy widely used in psychiatric practice， and the optimal first stimulus dosage in MECT is usually estimated clinically based on the factors influencing the patient's initial seizure threshold （IST）. However， previous studies on the influencing factors of IST have mostly suffered from limitations such as small sample sizes and single-dimensional research perspectives. ObjectiveTo explore the factors influencing IST in MECT for patients with mental disorders， so as to provide references for stimulus dosing strategies in MECT for the patients. MethodsA retrospective study was used to include 1 446 inpatients fulfilling the diagnostic criteria for any specific mental disorder listed in the ICD-10 and receiving MECT at Shandong Daizhuang Hospital from January 1， 2021 to August 1， 2023. Their general and clinical data were collected， including IST， psychiatric diagnostic categories， gender， ethnicity， age， body weight， body mass index （BMI）， course of disease， family history of psychiatric disorders， first episode status， use of antiepileptic drugs the day before treatment， use of benzodiazepines the day before treatment， and previous MECT treatment history. Pearson correlation analysis was utilized to test the correlation of IST with age， height， body weight， BMI， and course of disease， and stepwise multivariate linear regression analysis was performed to identify the factors affecting IST. ResultsIST yielded statistical difference among patients in terms of gender， first episode status， use of antiepileptic drugs the day before treatment， and use of benzodiazepines the day before treatment （t=2.256， -3.059， -2.136， -3.006， P<0.05 or 0.01）. IST in patients of different ages and psychiatric diagnostic categories also demonstrated statistical difference （F=913.120， 6.212， P<0.01）. Within young population， IST varied significantly based on the psychiatric diagnostic categories （F=2.986， P<0.05）. Correlation analysis indicated that IST was positively correlated with age， body weight， BMI and course of disease （r=0.886， 0.055， 0.184， 0.456， P<0.05 or 0.01）， and negatively correlated with height （r=-0.183， P<0.01）. Stepwise multivariate linear regression analysis revealed that age， gender， and body weight were influencing factors of IST （β=0.888， -0.049， -0.035， P<0.01）. ConclusionsAge， gender and body weight may be factors influencing IST in MECT for patients with mental disorders. ［Funded by Key R&D Plan Projects of Jining City in 2024 (number, 2024YXNS202）］

ObjectiveTo investigate the value of Fried Frailty Phenotype （FFP）， liver frailty index （LFI）， and Short Physical Performance Battery （SPPB） in predicting 2-year all-cause mortality and decompensation events in patients with liver cirrhosis. MethodsA total of 277 patients with liver cirrhosis who were hospitalized in Beijing Friendship Hospital， Capital Medical University， from December 2020 to December 2021 were enrolled， and FFP， LFI， and SPPB were used to assess the state of frailty. Based on the scores of each tool， these patients were divided into frail and non-frail groups. These three tools were compared in terms of consistency and independent predictive performance. The primary endpoints were 2-year all-cause mortality rate and composite endpoints （death+decompensation events）， and the Cox regression analysis， the receiver operating characteristic （ROC） curve， net reclassification index （NRI）， and integrated discrimination improvement （IDI） index were used to analyze the predictive value of the three tools. Normally distributed continuous data were compared between two groups using the independent samples t-test， while non-normally distributed continuous data were compared using the Mann-Whitney U test. Categorical data were compared between groups using the chi-square test or Fisher’s exact test. The agreement among different frailty tools was evaluated using Cohen’s Kappa statistic. The Kaplan-Meier survival curve was plotted， and a survival analysis was performed using the log-rank test. ResultsThe prevalence rate of frailty assessed by FFP， LFI， and SPPB was 37.2%， 22.4%， and 20.2%， respectively， with a moderate consistency between FFP and LFI/SPPB （κ=0.57， 95% confidence interval ［CI］： 0.47 — 0.67； κ=0.51， 95%CI： 0.41 — 0.62） and a relatively high consistency between LFI and SPPB （κ=0.87， 95%CI： 0.80 — 0.94）. Compared with the non-frailty group， the frailty group had significantly higher all-cause mortality rate and incidence rate of composite endpoints （P0.001）. After multivariate adjustment， FFP， LFI， and SPPB had a hazard ratio of 2.42（95%CI： 1.51 — 5.11）， 2.21（95%CI： 1.11 — 4.42）， and 2.21（95%CI： 1.14 — 4.30）， respectively， in predicting all-cause mortality， as well as a hazard ratio of 2.51（95%CI： 1.61 — 3.91）， 2.40（95%CI： 1.51 — 3.80）， and 2.20（95%CI： 1.39 — 3.47）， respectively， in predicting composite endpoints. Compared with Child-Pugh score， FFP had a significantly greater area under the ROC curve （AUC） in predicting all-cause mortality （0.79 vs 0.69， P=0.032） and composite endpoints （0.75 vs 0.68， P=0.044）. Frailty assessment tools combined with Child-Pugh score significantly improved the performance in predicting all-cause mortality and composite endpoints， with an AUC of 0.81 — 0.82 and 0.77 — 0.78， respectively （P0.05）. NRI and IDI analyses further confirmed the improvement of the combined model in classification （all P0.001）. ConclusionFFP， LFI， and SPPB can independently predict adverse outcomes in patients with liver cirrhosis， among which FFP has the best predictive performance， and the combination of frailty assessment tools with Child-Pugh score can significantly enhance the accuracy of prognostic evaluation.

Purpose: The genomic characteristics of uterine sarcomas have not been fully elucidated. This study aimed to explore the genomic landscape of the uterine sarcomas (USs). Materials and Methods: Comprehensive genomic analysis through RNA-sequencing was conducted. Gene fusion, differentially expressed genes (DEGs), signaling pathway enrichment, immune cell infiltration, and prognosis were analyzed. A deep learning model was constructed to predict the survival of US patients. Results: A total of 71 US samples were examined, including 47 endometrial stromal sarcomas (ESS), 18 uterine leiomyosarcomas (uLMS), three adenosarcomas, two carcinosarcomas, and one uterine tumor resembling an ovarian sex-cord tumor. ESS (including high-grade ESS [HGESS] and low-grade ESS [LGESS]) and uLMS showed distinct gene fusion signatures; a novel gene fusion site, MRPS18A–PDC-AS1 could be a potential diagnostic marker for the pathology differential diagnosis of uLMS and ESS; 797 and 477 uterine sarcoma DEGs (uDEGs) were identified in the ESS vs. uLMS and HGESS vs. LGESS groups, respectively. The uDEGs were enriched in multiple pathways. Fifteen genes including LAMB4 were confirmed with prognostic value in USs; immune infiltration analysis revealed the prognositic value of myeloid dendritic cells, plasmacytoid dendritic cells, natural killer cells, macrophage M1, monocytes and hematopoietic stem cells in USs; the deep learning model named Max-Mean Non-Local multi-instance learning (MMN-MIL) showed satisfactory performance in predicting the survival of US patients, with the area under the receiver operating curve curve reached 0.909 and accuracy achieved 0.804. Conclusion USs harbored distinct gene fusion characteristics and gene expression features between HGESS, LGESS, and uLMS. The MMN-MIL model could effectively predict the survival of US patients.