ObjectiveTo explore the molecular mechanism of Buzhong Yiqitang in attenuating cisplatin resistance of non-small cell lung cancer （NSCLC） cells （A549/DDP） by regulating endoplasmic reticulum stress （ERS） via the nuclear factor E₂-related factor 2 （Nrf2）/reactive oxygen species （ROS）/double-stranded RNA-activated protein kinase R （PKR）-like endoplasmic reticulum kinase （PERK）/CCAAT enhancer-binding protein homologous protein （CHOP） signaling pathway. MethodsSprague Dawley （SD） rats were used to prepare blank serum and Buzhong Yiqitang-containing serum samples， and A549/DDP cells were sub-cultured and randomly allocated into 9 groups： Blank （10% blank rat serum medium）， model （10% blank rat serum medium+20 mg·L^-1 cisplatin）， traditional Chinese medicine（TCM） （10% Buzhong Yiqitang-containing rat serum medium+20 mg·L^-1 cisplatin）， TBHQ （10% blank rat serum medium+5 μmol·L^-1 TBHQ+20 mg·L^-1 cisplatin）， TBHQ+TCM （10% Buzhong Yiqitang-containing rat serum medium+5 μmol·L^-1 TBHQ+20 mg·L^-1 cisplatin）， NAC （10% blank rat serum medium+600 μmol·L^-1 NAC+20 mg·L^-1 cisplatin）， NAC+TCM （10% Buzhong Yiqitang-containing rat serum medium+600 μmol·L^-1 NAC+20 mg·L^-1 cisplatin）， Salubrinal （10% blank rat serum medium+20 μmol·L^-1 Salubrinal+20 mg·L^-1 cisplatin）， and Salubrinal+TCM （10% Buzhong Yiqitang-containing rat serum medium+20 μmol·L^-1 Salubrinal+20 mg·L^-1 cisplatin）. The median inhibitory concentration （IC₅₀） of cisplatin in each group was determined by the cell counting kit-8 （CCK-8） method. The ROS level was detected by flow cytometry. The ultrastructure of endoplasmic reticulum （ER） was observed by transmission electron microscopy. Western blot was employed to determine the protein levels of Nrf2， phosphorylated （p）-Nrf2， CHOP， PERK， p-PERK， eukaryotic translation initiation factor 2α （eIF2α）， p-eIF2α， and activated transcription factor 4 （ATF4）. The fluorescence intensity of CHOP and p-PERK was detected by confocal fluorescence localization. ResultsCompared with the model group， the TCM group showed a decrease in IC₅₀ of cisplatin in A549/DDP cells （P<0.01）. TBHQ， NAC， and Salubrinal groups showed increases in IC₅₀ of cisplatin in A549/DDP cells （P<0.01）. TCM combined with TBHQ， NAC， and Salubrinal reduced the IC₅₀ （P<0.01）. Compared with that in the blank group， the ROS level in the model group elevated （P<0.01）. Compared with that in the model group， the ROS level in the TCM group elevated （P<0.01）. The ROS levels in the TBHQ， NAC， and Salubrinal groups were lower than that in the model group （P<0.01）. The combinations of TBHQ， NAC， and Salubrinal with TCM raised the ROS levels of A549/DDP cells （P<0.01）. A flat saclike and neatly arranged ER structure was observed in the blank group， and slight expansion of ER was observed in the model group. Significant expansion of ER was observed in the TCM group. The expansion of ER was not obvious in the TBHQ， NAC， and Salubrinal groups but significant after TBHQ， NAC， and Salubrinal were combined with TCM. CHOP and p-PERK showed higher fluorescence intensity in the model group than in the blank group （P<0.01）， and the fluorescence signal intensity in the TCM group was higher than that in the model group （P<0.01）. The fluorescence intensity in the TBHQ， NAC， and Salubrinal groups was lower than that in the model group （P<0.01）. The fluorescence intensity in the groups of TBHQ， NAC， and Salubrinal combined with TCM was higher than that in corresponding groups without TCM （P<0.01）. Compared with those in the blank group， the expression levels of Nrf2 and p-Nrf2 were up-regulated in the model group （P<0.05）. Compared with the model group， the TCM group showed down-regulated expression levels of Nrf2 and p-Nrf2 （P<0.05）. TBHQ， NAC， and Salubrinal increased the expression levels of Nrf2 and p-Nrf2 （P<0.05）， while their combinations with TCM decreased the expression levels of Nrf2 and p-Nrf2 compared with the corresponding groups without TCM （P<0.01）. There were no significant differences in the expression levels of PERK and eIF2α between groups. The expression levels of CHOP， p-PERK/PERK， p-eIF2α/eIF2α， and ATF4 were up-regulated in the model group compared with those in the blank group （P<0.05）. Compared with the model group， the TCM group showed up-regulated expression levels of CHOP， p-PERK/PERK， p-eIF2α/eIF2α， and ATF4 （P<0.05）， while the TBHQ， NAC， and Salubrinal groups showed down-regulated expression levels （P<0.05）. The groups of TBHQ， NAC， and Salubrinal combined with TCM had higher expression levels than the groups without TCM （P<0.01）. ConclusionBuzhong Yiqitang regulates ERS via the Nrf2/ROS/PERK/CHOP signaling pathway to attenuate cisplatin resistance in NSCLC.

ObjectiveTo explore the molecular mechanism of Buzhong Yiqitang in attenuating cisplatin resistance in non-small cell lung cancer （NSCLC） by inducing ferroptosis via poly（rC）-binding protein 1 （PCBP1）. MethodsThe serum containing Buzhong Yiqitang was prepared and cisplatin-resistant human non-small cell lung cancer （NSCLC） cells （A549/DDP） were cultured and randomly grouped as follows： Blank （10% blank serum）， model （10% blank serum+20 mg·L^-1 cisplatin）， Buzhong Yiqitang （10% serum containing Buzhong Yiqitang+20 mg·L^-1 cisplatin）， Fe-1 （10% blank serum+20 mg·L^-1 cisplatin+5 μmol·L^-1 Fe-1）， and Buzhong Yiqitang+Fe-1 （10% serum containing Buzhong Yiqitang+20 mg·L^-1 cisplatin+5 μmol·L^-1 Fe-1）. Firstly， PCR Array was used to screen ferroptosis-related genes regulated by Buzhong Yiqitang， and PCBP1 was identified as the target for studying the attenuation of cisplatin resistance by Buzhong Yiqitang. Subsequently， the median inhibitory concentration （IC₅₀） of cisplatin in each group was determined by the cell counting kit-8 （CCK-8） method and the resistance index （RI） was calculated. The ultrastructure of A549/DDP cells in each group was observed by transmission electron microscopy. The protein levels of PCBP1 and glutathione peroxidase 4 （GPX4） were determined by Western blot. The lipid reactive oxygen species （ROS） content in each group was determined by the C11-BODIRY 581/591 fluorescence probe. The ferrous ion assay kit was used to measure the ferrous ion content in each group. The malondialdehyde （MDA） assay kit was used to determine the MDA content in each group. ResultsCompared with model group， the IC₅₀ of cisplatin and the RI of A549/DDP cells decreased in the Buzhong Yiqitang group （P<0.05） but increased in the Fe-1 group （P<0.05）. The IC₅₀ of cisplatin and the RI of A549/DDP cells in the Buzhong Yiqitang+Fe-1 group were lower than those in the Fe-1 group （P<0.05）. Compared with the model group， the Buzhong Yiqitang group showed obvious mitochondrial ferroptosis， while the mitochondrial damage became less obvious after Fe-1 treatment. Compared with that in the Fe-1 group， the mitochondrial ferroptosis was aggravated after the intervention with Buzhong Yiqitang. Compared with blank group， the model group showed down-regulated expression levels of PCBP1 and GPX4 （P<0.05） and increased content of lipid ROS， ferrous ions， and MDA （P<0.05） in A549/DDP cells. Compared with model group， the Buzhong Yiqitang group showed down-regulated expression levels of PCBP1 and GPX4 （P<0.05） and increased content of lipid ROS， ferrous ions， and MDA （P<0.05）， while the Fe-1 group showed up-regulated expression levels of PCBP1 and GPX4 （P<0.05） and reduced content of lipid ROS， ferrous ions， and MDA （P<0.05）. Compared with the Fe-1 group， the Buzhong Yiqitang+Fe-1 group showed down-regulated expression levels of PCBP1 and GPX4 and increased content of lipid ROS， ferrous ions， and MDA （P<0.05）. ConclusionBuzhong Yiqitang attenuated cisplatin resistance in NSCLC by regulating PCBP1 to induce ferroptosis.

In the process of tumor chemotherapy， the emergence of multi-drug resistance （MDR） has always been a thorny problem， which is a result of the joint action of the host， tumor cells， and the immune microenvironment. Tumor cells can escape the toxicity of chemotherapeutic drugs through multiple pathways， being easy to produce drug resistance. MDR greatly restricts the effect of chemotherapeutic drugs on tumor cells and affects their therapeutic effects. Traditional Chinese medicine （TCM） has the unique advantages of multi-target， multi-pathway and individualized treatment. The TCM treatment of tumors emphasizes regulating Yin and Yang， as well as reinforcing healthy Qi and dispelling pathogen. In recent years， TCM has demonstrated remarkable efficacy in the treatment of tumors and the amelioration of multi-drug resistance. TCM not only can target the phenomenon of MDR but also greatly weakens the side effects of the patients after the chemotherapy， thus improving the survival quality and rate of the patients. Accordingly， many patients adopt TCM as an adjuvant therapy during or after chemotherapy. The binding of TCM to targets can reverse the drug resistance of various tumors， which has become an emerging research highlight. From the regulatory mechanism of TCM on MDR of tumors， this paper introduces the mechanisms by which tumor cells continue to grow， proliferate， and metastasize by adjusting the intracellular drug concentration， altering or utilizing the tumor microenvironment， and affecting the cell death mode to achieve the resistance to chemotherapeutic drugs. In this regard， the active ingredients and compound prescriptions of TCM can increase the sensitivity of chemotherapeutic drugs by down-regulating drug transporters， improving the tumor microenvironment， and modulating the drug resistance pathways associated with apoptosis， autophagy， ferroptosis， or pyroptosis. The aim of this paper is to explore more clinical practical value of TCM in the treatment of tumors and provide exploratory ideas and a theoretical basis for the future research on tumors and MDR.

Objective:To analyze the clinical characteristics of children with head and neck rhabdomyosarcoma (RMS) and to summarize the mid-long term efficacy of Beijing Children′s Hospital Rhabdomyosarcoma 2006 (BCH-RMS-2006) regimen and China Children′s Cancer Group Rhabdomyosarcoma 2016 (CCCG-RMS-2016) regimen.Methods:A retrospective cohort study. Clinical data of 137 children with newly diagnosed head and neck RMS at Beijing Children′s Hospital, Capital Medical University from March 2013 to December 2021 were collected. Clinical characteristic of patients at disease onset and the therapeutic effects of patients treated with the BCH-RMS-2006 and CCCG-RMS-2016 regimens were compared. The treatments and outcomes of patients with recurrence were also summarized. Survival analysis was performed by Kaplan-Meier method, and Log-Rank test was used for comparison of survival rates between groups.Results:Among 137 patients, there were 80 males (58.4%) and 57 females (41.6%), the age of disease onset was 59 (34, 97) months. The primary site in the orbital, non-orbital non-parameningeal, and parameningeal area were 10 (7.3%), 47 (34.3%), and 80 (58.4%), respectively. Of all patients, 32 cases (23.4%) were treated with the BCH-RMS-2006 regimen and 105 (76.6%) cases were treated with the CCCG-RMS-2016 regimen. The follow-up time for the whole patients was 46 (20, 72) months, and the 5-year progression free survival (PFS) and overall survival (OS) rates for the whole children were (60.4±4.4)% and (69.3±4.0)%, respectively. The 5-year OS rate was higher in the CCCG-RMS-2016 group than in BCH-RMS-2006 group ((73.0±4.5)% vs. (56.6±4.4)%, χ2=4.57, P=0.029). For the parameningeal group, the 5-year OS rate was higher in the CCCG-RMS-2016 group (61 cases) than in BCH-RMS-2006 group (19 cases) ((57.3±7.6)% vs. (32.7±11.8)%, χ2=4.64, P=0.031). For the group with meningeal invasion risk factors, the 5-year OS rate was higher in the CCCG-RMS-2016 group (54 cases) than in BCH-RMS-2006 group (15 cases) ((57.7±7.7)% vs. (30.0±12.3)%, χ2=4.76, P=0.029). Among the 10 cases of orbital RMS, there was no recurrence. In the non-orbital non-parameningeal RMS group (47 cases), there were 13 (27.6%) recurrences, after re-treatment, 7 cases survived. In the parameningeal RMS group (80 cases), there were 40 (50.0%) recurrences, with only 7 cases surviving after re-treatment. Conclusions:The overall prognosis for patients with orbital and non-orbital non-parameningeal RMS is good. However, children with parameningeal RMS have a high recurrence rate, and the effectiveness of re-treatment after recurrence is poor. Compared with the BCH-RMS-2006 regimen, the CCCG-RMS-2016 regimen can improve the treatment efficacy of RMS in the meningeal region.

Objective:To analyze factors influencing repigmentation patterns in patients with vitiligo treated with phototherapy.Methods:Clinical data were retrospectively collected from patients with vitiligo treated with 308-nm excimer laser or 308-nm excimer lamp at the Department of Dermatology, Xijing Hospital, Air Force Medical University from June 2013 to May 2022. The treatment frequency was thrice weekly, and skin lesions were evaluated via photographs once every 5 sessions of phototherapy. Chi-square test or Fisher′s exact test was used to analyze associations between clinical characteristics and vitiligo repigmentation patterns.Results:A total of 223 patients with vitiligo were included in this study, including 109 males (48.9%) and 114 females (51.1%), and their ages ( M [ Q1, Q3]) were 20 (10, 28) years. Among the 223 patients, 170 (76.2%) were treated with 308-nm excimer laser, and 53 (23.8%) with 308-nm excimer lamp. The repigmentation patterns included the perifollicular pattern in 63 cases (28.3%), marginal pattern in 97 (43.5%), diffuse pattern in 36 (16.1%), and mixed pattern in 27 (12.1%). Analysis of the associations between clinical characteristics and vitiligo repigmentation patterns showed no significant differences in the repigmentation patterns among vitiligo patients of different genders or different Fitzpatrick skin types (both P > 0.05) ; however, the diffuse repigmentation pattern more frequently occurred in the patients aged ≤ 12 years compared with those aged > 12 years ( χ2 = 7.71, P = 0.005), in the patients with vitiligo in the progressive stage compared with those in the stable stage ( χ2 = 4.59, P = 0.030), and in lesions without white hair compared with those with white hair ( χ2 = 6.75, P = 0.009) ; the mixed repigmentation pattern more frequently occurred in the patients with segmental vitiligo compared with those with non-segmental vitiligo ( χ2 = 11.76, P = 0.001) ; the marginal repigmentation pattern more frequently occurred in lesions on the face and neck ( χ2 = 15.82, P<0.001) and extremities ( χ2 = 11.85, P = 0.001) compared with lesions on the trunk; the perifollicular repigmentation pattern more frequently occurred in the patients with stable vitiligo compared with those with progressive vitiligo ( χ2 = 4.70, P = 0.030), and in skin lesions on the trunk compared with those on face and neck ( χ2 = 13.73, P < 0.001) and extremities ( χ2 = 5.49, P = 0.035) ; after 308-nm excimer laser treatment, the proportions of patients with the marginal repigmentation pattern ( χ2 = 12.30, P < 0.001) and those with the diffuse repigmentation pattern ( χ2 = 5.64, P = 0.018) were significantly higher than those after 308-nm excimer lamp treatment, while the proportions of patients with the perifollicular repigmentation pattern ( χ2 = 7.87, P = 0.005) and those with the mixed repigmentation pattern ( χ2 = 17.13, P < 0.001) were significantly higher after 308-nm excimer lamp treatment than those after 308-nm excimer laser treatment. Conclusion:Patients′ age, clinical types and stages of vitiligo, presence or absence of concomitant white hair, skin lesion sites, and phototherapy modalities were factors influencing the repigmentation patterns of vitiligo.

ObjectiveTo explore the mechanism of Buzhong Yiqitang-containing serum in alleviating the cisplatin resistance in human non-small cell lung cancer （A549/DDP） cells via regulating the nuclear factor E₂-related factor 2 （Nrf2）/reactive oxygen species （ROS） signaling pathway. MethodThe serum containing Buzhong Yiqitang was prepared and A549/DDP cells were cultured and randomly grouped： blank （10% blank serum）， cisplatin （10% blank serum+20 mg·L^-1 cisplatin）， Buzhong Yiqitang （10% Buzhong Yiqitang-containing serum+20 mg·L^-1 cisplatin）， ML385 （10% blank serum+5 μmol·L^-1 ML385+20 mg·L^-1 cisplatin）， Buzhong Yiqitang+ML385 （10% Buzhong Yiqitang-containing serum+5 μmol·L^-1 ML385+20 mg·L^-1 cisplatin）， tertiary butylhydroquinone （TBHQ） （10% blank serum+5 μmol·L^-1 TBHQ+20 mg·L^-1 cisplatin）， and Buzhong Yiqitang+TBHQ （10% Buzhong Yiqitang-containing serum+5 μmol·L^-1 TBHQ+20 mg·L^-1 cisplatin）. The median inhibitory concentration （IC₅₀） of cisplatin in each group was determined by the cell counting kit-8 （CCK-8） method and the resistance index （RI） was calculated. The apoptosis rate was detected by flow cytometry. The ROS content of each group was determined with the DCFH-DA fluorescence probe. Western blot was employed to determine the protein levels of Nrf2， cleaved cysteinyl aspartate-specific protease-3 （cleaved Caspase-3）， cytochrome C （Cyt C）， and B-cell lymphoma-2 （Bcl-2）. ResultCompared with those in the cisplatin group， the IC₅₀ and RI of A549/DDP cells to cisplatin in Buzhong Yiqitang， ML385， and Buzhong Yiqitang+ML385 groups decreased （P˂0.05）. Compared with the blank group， the cisplatin， Buzhong Yiqitang， ML385， and Buzhong Yiqitang+ML385 groups showed increased apoptosis rate of A549/DDP cells （P˂0.05）. Compared with the blank group， cisplatin promoted the expression of Nrf2 （P˂0.05）. Compared with the cisplatin group， Buzhong Yiqitang， ML385， and Buzhong Yiqitang+ML385 inhibited the expression of Nrf2 （P<0.05）， elevated the ROS level （P˂0.05）， up-regulated the protein levels of cleaved Caspase-3 and Cyt C， and down-regulated the protein level of Bcl-2 （P<0.05）， which were the most significant in the Buzhong Yiqitang+ML385 group. Compared with the cisplatin group， the TBHQ group showed increased IC₅₀ and RI of cisplatin （P<0.05）， decreased apoptosis rate of A549/DDP cells （P<0.05）， up-regulated protein levels of Nrf2 and Bcl-2 （P<0.05）， lowered level of ROS （P˂0.05）， and down-regulated protein levels of cleaved Caspase-3 and Cyt C （P<0.05）. Compared with the TBHQ group， Buzhong Yiqitang+TBHQ decreased the IC₅₀ and RI of cisplatin in A549/DDP cells （P<0.05）， increased the apoptosis rate （P<0.05）， down-regulated the protein levels of Nrf2 and Bcl-2 （P<0.05）， increased ROS （P˂0.05）， and up-regulated the protein levels of cleaved Caspase-3 and Cyt C （P<0.05）. ConclusionBuzhong Yiqitang induced apoptosis by inhibiting Nrf2/ROS pathway to alleviate cisplatin resistance in A549/DDP cells.

Objective To investigate the clinical diagnostic value of plasma serine protease inhibitor Ka-zal-type 4(SPINK4)expression in colorectal adenocarcinoma(CRC)and progressive adenoma(AA).Methods A total of 62 patients with CRC(CRC group)and 15 patients with AA(AA group)diagnosed by colonoscopy and pathological examination in this hospital from June 2020 to December 2021 were selected,and 22 healthy people undergoing physical examination during the same period were selected as the HC group.The expression of SPINK4 in plasma was detected by ELISA,and the expression of CEA in plasma was detected by electrochemiluminescence,and the correlation was analyzed.The diagnostic efficiency was analyzed by re-ceiver operating characteristic(ROC)curve,and the expression of p53 in CRC tissues was detected by immu-nohistochemistry.Results The expression of plasma SPINK4 in the CRC group and AA group was lower than that in the HC group(Z=3.72,-0.41,P＜0.05),and the expression of CEA in the CRC group was higher than that in the HC group(Z=-3.63,P＜0.05).The area under the curve(AUC),accuracy,sensi-tivity and specificity of SPINK4 combined with CEA in the diagnosis of CRC and AA were higher than those of SPINK4 and CEA alone.The positive rate of mutant type p53 in SPINK4 low expression group and CEA high ex-pression group was significantly increased in CRC patients(72.55％,75.00％,P＜0.05).Conclusion The expression of plasma SPINK4 is decreased in CRC and AA,and the combined detection of SPINK4 and CEA has a good di-agnostic efficiency in CRC and AA.

Vitiligo is an immune memory skin disease. T-cell factor 1 (TCF1) is essential for maintaining the memory T-cell pool. There is an urgent need to investigate the characteristics of peripheral memory T-cell profile and TCF1⁺ T-cell frequencies in patients with vitiligo. In this study, 31 patients with active vitiligo (AV), 22 with stable vitiligo (SV), and 30 healthy controls (HCs) were included. We measured circulating memory and TCF1⁺ T-cell frequencies using flow cytometry. The Spearman's rank test was used to evaluate the correlation between cell frequencies and disease characteristics. Receiver operating characteristic curves (ROC) were constructed to investigate the discriminative power of the cell subpopulations. Circulating CD4⁺ and CD8⁺ terminally differentiated effector memory T-cell (T_EMRA) frequencies were significantly higher in the AV group than in HCs (P < 0.05). TCF1⁺ T-cell subpopulations were widespread increased in patients with vitiligo (P < 0.05). After adjusting for potential confounders, CD8⁺ and CD4⁺ central memory (T_CM) cells, and CD8⁺ T_EMRA were correlated with disease activity (P < 0.05). The combined diagnostic value of the four (naïve, effector memory, T_CM, and T_EMRA) CD8⁺TCF1⁺ T-cell subsets was relatively high (area under the ROC curve (AUC) = 0.804, sensitivity = 71.70%, specificity = 83.34%), and the CD8⁺ T-cell subsets combination performed well in discriminating disease activity (AUC = 0.849, sensitivity = 70.97%, specificity = 90.91%). We demonstrated an altered circulating memory T-cell profile and increased TCF1⁺ T-cell percentage in patients with vitiligo. T-cell subpopulations had a strong value for vitiligo diagnosis and activity evaluation. This evidence presents a potential new pharmacological target for inhibiting autoimmunity that leads to vitiligo.

Objective: On the basis of the dominant frequency index of functional connectivity, the "brain age" analysis method was used to explore abnormal development patterns of sensorimotor networks in boys with autism spectrum disorder(ASD). Methods: The resting state functional magnetic resonance data (7-12 years old) for 105 boys with ASD and 102 matched boys with normal development from the ABIDE public database were screened. Functional connection networks in different frequency bands of sensorimotor related brain regions were constructed for each individual, and the frequency of the strongest connection were constructed as the optimal frequency of the connection. Brain age analysis was used to explore the difference between brain age and chronological age in boys with ASD. Results: The brain sensorimotor network of boys with ASD showed an abnormal development pattern of overdevelopment followed by underdevelopment, and the transition between the two patterns occurred at approximately 7.8 years of age. Older boys with ASD (older than 10 years) whose underdevelopment trend was suppressed had lower ASD severity( r=-0.43, P < 0.05 ). Conclusion The brain sensorimotor network in boys with ASD has an abnormal development process, and the brain chronological age difference in the sensorimotor network has potential as a neuroimaging marker to measure the development of ASD.

OBJECTIVE: To analyze the risk factors of in-hospital death in patients with sepsis in the intensive care unit (ICU) based on machine learning, and to construct a predictive model, and to explore the predictive value of the predictive model. METHODS: The clinical data of patients with sepsis who were hospitalized in the ICU of the Affiliated Hospital of Jining Medical University from April 2015 to April 2021 were retrospectively analyzed,including demographic information, vital signs, complications, laboratory examination indicators, diagnosis, treatment, etc. Patients were divided into death group and survival group according to whether in-hospital death occurred. The cases in the dataset (70%) were randomly selected as the training set for building the model, and the remaining 30% of the cases were used as the validation set. Based on seven machine learning models including logistic regression (LR), K-nearest neighbor (KNN), support vector machine (SVM), decision tree (DT), random forest (RF), extreme gradient boosting (XGBoost) and artificial neural network (ANN), a prediction model for in-hospital mortality of sepsis patients was constructed. The receiver operator characteristic curve (ROC curve), calibration curve and decision curve analysis (DCA) were used to evaluate the predictive performance of the seven models from the aspects of identification, calibration and clinical application, respectively. In addition, the predictive model based on machine learning was compared with the sequential organ failure assessment (SOFA) and acute physiology and chronic health evaluation II (APACHE II) models. RESULTS: A total of 741 patients with sepsis were included, of which 390 were discharged after improvement, 351 died in hospital, and the in-hospital mortality was 47.4%. There were significant differences in gender, age, APACHE II score, SOFA score, Glasgow coma score (GCS), heart rate, oxygen index (PaO₂/FiO₂), mechanical ventilation ratio, mechanical ventilation time, proportion of norepinephrine (NE) used, maximum NE, lactic acid (Lac), activated partial thromboplastin time (APTT), albumin (ALB), serum creatinine (SCr), blood urea nitrogen (BUN), blood uric acid (BUA), pH value, base excess (BE), and K⁺ between the death group and the survival group. ROC curve analysis showed that the area under the curve (AUC) of RF, XGBoost, LR, ANN, DT, SVM, KNN models, SOFA score, and APACHE II score for predicting in-hospital mortality of sepsis patients were 0.871, 0.846, 0.751, 0.747, 0.677, 0.657, 0.555, 0.749 and 0.760, respectively. Among all the models, the RF model had the highest precision (0.750), accuracy (0.785), recall (0.773), and F1 score (0.761), and best discrimination. The calibration curve showed that the RF model performed best among the seven machine learning models. DCA curve showed that the RF model exhibited greater net benefit as well as threshold probability compared to other models, indicating that the RF model was the best model with good clinical utility. CONCLUSIONS The machine learning model can be used as a reliable tool for predicting in-hospital mortality in sepsis patients. RF models has the best predictive performance, which is helpful for clinicians to identify high-risk patients and implement early intervention to reduce mortality.
Humans ; Hospital Mortality ; Retrospective Studies ; ROC Curve ; Prognosis ; Sepsis/diagnosis* ; Intensive Care Units