PURPOSE: Septic shock is associated with high mortality and poor outcomes among sepsis patients with coagulopathy. Although traditional statistical methods or machine learning (ML) algorithms have been proposed to predict septic shock, these potential approaches have never been systematically compared. The present work aimed to develop and compare models to predict septic shock among patients with sepsis. METHODS: It is a retrospective cohort study based on 484 patients with sepsis who were admitted to our intensive care units between May 2018 and November 2022. Patients from the 908th Hospital of Chinese PLA Logistical Support Force and Nanchang Hongdu Hospital of Traditional Chinese Medicine were respectively allocated to training (n=311) and validation (n=173) sets. All clinical and laboratory data of sepsis patients characterized by comprehensive coagulation indexes were collected. We developed 5 models based on ML algorithms and 1 model based on a traditional statistical method to predict septic shock in the training cohort. The performance of all models was assessed using the area under the receiver operating characteristic curve and calibration plots. Decision curve analysis was used to evaluate the net benefit of the models. The validation set was applied to verify the predictive accuracy of the models. This study also used Shapley additive explanations method to assess variable importance and explain the prediction made by a ML algorithm. RESULTS: Among all patients, 37.2% experienced septic shock. The characteristic curves of the 6 models ranged from 0.833 to 0.962 and 0.630 to 0.744 in the training and validation sets, respectively. The model with the best prediction performance was based on the support vector machine (SVM) algorithm, which was constructed by age, tissue plasminogen activator-inhibitor complex, prothrombin time, international normalized ratio, white blood cells, and platelet counts. The SVM model showed good calibration and discrimination and a greater net benefit in decision curve analysis. CONCLUSION The SVM algorithm may be superior to other ML and traditional statistical algorithms for predicting septic shock. Physicians can better understand the reliability of the predictive model by Shapley additive explanations value analysis.
Humans ; Shock, Septic/blood* ; Machine Learning ; Male ; Female ; Retrospective Studies ; Middle Aged ; Aged ; Sepsis/complications* ; ROC Curve ; Cohort Studies ; Adult ; Intensive Care Units ; Algorithms ; Blood Coagulation ; Critical Illness

Objective To explore the risk factors for noncontiguous spinal fractures(NSFs)in the elderly.Methods The clinical data of 614 elderly patients with spinal fracture from January 2013 to December 2019 were analyzed retrospectively.Patients were divided into the NSFs group and the Non-NSFs group according to whether NSFs occurred or not.Univariate analysis and multivariate Logistic regression analysis were used to screen the risk factors of NSFs.Results Univariate analysis showed that female(P=0.003),high-energy violent injury(P=0.032),osteoporosis(P=0.004),fracture in spring(P=0.020),and previous spinal fracture history(P<0.001)were associated with the occurrence of NSFs.Multivariate Logistic regression analysis showed that fracture in spring(P=0.024),previous spinal fracture history(P<0.001)and high-energy violent injury(P=0.038)were the independent risk factors for the occurrence of NSFs in the elderly.Conclusion High-energy violent injury,fracture in spring and previous spinal fracture history are the independent risk factors for the occurrence of NSFs in the elderly.Therefore,elderly patients with the above risk factors should be examined more carefully and comprehensively to avoid missed diagnosis and delayed diagnosis.In order to reduce the incidence of this disease,corresponding measures should be taken according to the preventable risk factors.

Objective To analyze the risk factors for increased drainage volume after open transforaminal lumbar interbody fusion(TLIF),and to establish a predictive model and then validate it.Methods The clinical data of 680 patients who underwent open TLIF at the General Hospital of Northern Theater Command from January 2016 to December 2019 were collected and the patients were randomly divided into the training group(n=476)and the validation group(n=204).Taking the predictive factors screened out by LASSO regression analysis as independent variables,a multivariate Logistic regression predictive model was constructed.The model was internally validated through the receiver operating characteristic(ROC)curve,Hosmer-Lemeshow goodness-of-fit test,and calibration curve,and its clinical utility was assessed via decision curve analysis(DCA).Results LASSO regression analysis screened out four predictive variables:age,number of surgical segments,operative duration,and intraoperative blood loss.The multivariate Logistic regression predictive model demonstrated that age≥60 years,number of surgical segments≥4,operative duration≥2 hours,and intraoperative blood loss≥200 mL were independent influencing factors for the increased postoperative drainage volume in patients undergoing TLIF(P<0.05).ROC curve analysis revealed an area under the curve(AUC)of 0.816(95%CI:0.798 to 0.867)in the training group and 0.783(95%CI:0.685 to 0.823)in the validation group,indicating that the predictive model had good discriminatory ability.Additionally,the Hosmer-Lemeshow goodness-of-fit test and calibration curve indicated that the predictive model had a good degree of fit,and the predicted probability was basically consistent with the actual probability,demonstrating a good calibration.The DCA results confirmed that this predictive model could be applied in clinical practice.Conclusion The risk factors for increased drainage volume after open TLIF include age,number of surgical segments,operative duration,and intraoperative blood loss.The predictive model established based on these factors demonstrates good performance,and it can be applied in clinical guidance for the selection of drainage tube removal time after TLIF.

Objective To analyze the risk factors for increased drainage volume after open transforaminal lumbar interbody fusion(TLIF),and to establish a predictive model and then validate it.Methods The clinical data of 680 patients who underwent open TLIF at the General Hospital of Northern Theater Command from January 2016 to December 2019 were collected and the patients were randomly divided into the training group(n=476)and the validation group(n=204).Taking the predictive factors screened out by LASSO regression analysis as independent variables,a multivariate Logistic regression predictive model was constructed.The model was internally validated through the receiver operating characteristic(ROC)curve,Hosmer-Lemeshow goodness-of-fit test,and calibration curve,and its clinical utility was assessed via decision curve analysis(DCA).Results LASSO regression analysis screened out four predictive variables:age,number of surgical segments,operative duration,and intraoperative blood loss.The multivariate Logistic regression predictive model demonstrated that age≥60 years,number of surgical segments≥4,operative duration≥2 hours,and intraoperative blood loss≥200 mL were independent influencing factors for the increased postoperative drainage volume in patients undergoing TLIF(P<0.05).ROC curve analysis revealed an area under the curve(AUC)of 0.816(95%CI:0.798 to 0.867)in the training group and 0.783(95%CI:0.685 to 0.823)in the validation group,indicating that the predictive model had good discriminatory ability.Additionally,the Hosmer-Lemeshow goodness-of-fit test and calibration curve indicated that the predictive model had a good degree of fit,and the predicted probability was basically consistent with the actual probability,demonstrating a good calibration.The DCA results confirmed that this predictive model could be applied in clinical practice.Conclusion The risk factors for increased drainage volume after open TLIF include age,number of surgical segments,operative duration,and intraoperative blood loss.The predictive model established based on these factors demonstrates good performance,and it can be applied in clinical guidance for the selection of drainage tube removal time after TLIF.

Objective To explore the risk factors for noncontiguous spinal fractures(NSFs)in the elderly.Methods The clinical data of 614 elderly patients with spinal fracture from January 2013 to December 2019 were analyzed retrospectively.Patients were divided into the NSFs group and the Non-NSFs group according to whether NSFs occurred or not.Univariate analysis and multivariate Logistic regression analysis were used to screen the risk factors of NSFs.Results Univariate analysis showed that female(P=0.003),high-energy violent injury(P=0.032),osteoporosis(P=0.004),fracture in spring(P=0.020),and previous spinal fracture history(P<0.001)were associated with the occurrence of NSFs.Multivariate Logistic regression analysis showed that fracture in spring(P=0.024),previous spinal fracture history(P<0.001)and high-energy violent injury(P=0.038)were the independent risk factors for the occurrence of NSFs in the elderly.Conclusion High-energy violent injury,fracture in spring and previous spinal fracture history are the independent risk factors for the occurrence of NSFs in the elderly.Therefore,elderly patients with the above risk factors should be examined more carefully and comprehensively to avoid missed diagnosis and delayed diagnosis.In order to reduce the incidence of this disease,corresponding measures should be taken according to the preventable risk factors.

From the perspective of the physiological basis of liver and kidney sharing the common source in traditional Chinese medicine(TCM),and by integrating the theory of kidney dominating bone,liver dominating tendon,and meridian sinew of TCM as well as the bone resorption and collapse theory,and non-uniform settlement theory and lower-limb musculoskeletal bowstring structure theory of modern orthopedics,the pathogenesis of osteonecrosis of the femoral head(ONFH)under the system of non-uniform settlement during bone resorption and multidimensional composite bowstring working in coordination with the theory of liver-kidney and muscle-bone was explored.The key to the TCM pathogenesis of ONFH lies in the deficiency of the liver and kidney,and then the imbalance of kidney yin-yang leads to the disruption of the dynamic balance of bone formation and bone resorption mediated by osteoblasts-osteoclasts,which manifests as the elevated level of bone metabolism and the enhancement of focal bone resorption in the femoral head,and then leads to the necrosis and collapse of the femoral head.It is considered that the kidney dominates bone,liver dominates tendon,and the tendon and bone together constitute the muscle-bone-joint dynamic and static system of the hip joint.The appearance of collapse destroys the originally balanced muscle-bone-joint system.Moreover,the failure of liver blood in the nourishment of muscles and tendons further exacerbates the imbalance of the soft tissues around the hip joint,accelerates the collapse of the muscle-bone-joint dynamic and static system,speeds up the process of femoral head collapse,and ultimately results in irreversible outcomes.Based on the above pathogenesis,the systematic integrative treatment of ONFH should be based on the TCM holistic concept,focuses on the focal improvement of internal and external blood circulation of the femoral head by various approaches,so as to rebuild the coordination of joint function.Moreover,attention should be paid to the physical constitution of the patients,and therapy of tonifying the kidney and regulating the liver can be used to restore the balance between osteogenesis and osteoblastogenesis,and to reconstruct the muscle-bone-joint system,so as to effectively delay or even prevent the occurrence of ONFH.

Objective To explore the efficiency of artificial intelligence and radiomics-assisted X-ray in diagnosis of lumbar osteoporotic vertebral compression fractures(OVCF).Methods The clinical data of 455 patients diagnosed as lumbar OVCF by MRI in our hospital were selected.The patients were divided into the training group(n=364)and the validation group(n=91),X-ray films were extracted,the image delineation,feature extraction and data analysis were carried out,and the artificial intelligence radiomics deep learning was applied to establish a diagnostic model for OVCF.After verifying the effectiveness of the model by receiver operating characteristic(ROC)curve,area under the curve(AUC),calibration curve,and decision curve analysis(DCA),the efficiencies of manual reading,model reading,and model-assisted manual reading of X-ray in the early diagnosis of OVCF were compared.Results The ROC curve,AUC and calibration curve proved that the model had good discrimination and calibration,and excellent diagnostic performance.DCA demonstrated that the model had a higher clinical net benefit.The diagnostic efficiency of the manual reading group:the accuracy rate was 0.89,the recall rate was 0.62.The diagnostic efficiency of the model reading group:the accuracy rate was 0.93,the recall rate was 0.86,the model diagnosis showed good predictive performance,which was significantly better than the manual reading group.The diagnostic efficiency of the model-assisted manual reading group:the accuracy rate was 0.92,the recall rate was 0.72,and the recall rate of the model-assisted manual reading group was higher than that of the manual reading group,but lower than that of the model reading group,indicating the superiority of the model diagnosis.Conclusion The diagnostic model established based on artificial intelligence and radiomics in this study has reached an ideal level of efficacy,with better diagnostic efficacy compared with manual reading,and can be used to assist X-ray in the early diagnosis of OVCF.

Objective To develop a clinical prediction model for predicting risk factors for prolonged hospital stay after anterior cervical discectomy and fusion(ACDF).Methods The clinical data of 914 patients underwent ACDF treatment for cervical spondylotic myelopathy(CSM)were retrospectively analyzed.According to the screening criteria,800 eligible patients were eventually included,and the patients were divided into the development cohort(n=560)and the validation cohort(n=240).LASSO regression was used to screen variables,and multivariate Logistic regression analysis was used to establish a prediction model.The prediction model was evaluated from three aspects:differentiation,calibration and clinical effectiveness.The performance of the model was evaluated by area under the curve(AUC)and Hosmer-Lemeshow test.Decision curve analysis(DCA)was used to evaluate the clinical effectiveness of the model.Results In this study,the five factors that were significantly associated with prolonged hospital stay were male,abnormal BMI,mild-to-moderate anemia,stage of surgery(morning,afternoon,evening),and alcohol consumption history.The AUC of the development cohort was 0.778(95%CI:0.740 to 0.816),with a cutoff value of 0.337,and that of the validation cohort was 0.748(95%CI:0.687 to 0.809),with a cutoff value of 0.169,indicating that the prediction model had good differentiation.At the same time,the Hosmer-Lemeshow test showed that the model had a good calibration degree,and the DCA proved that it was effective in clinical application.Conclusion The prediction model established in this study has excellent comprehensive performance,which can better predict the risk of prolonged hospital stay,and can guide clinical intervention as soon as possible,so as to minimize the postoperative hospital stay and reduce the cost of hospitalization.

Objective To explore the risk factors for surgical site infection(SSI)after transforaminal lumbar interbody fusion(TLIF)for the treatment of lumbar degenerative diseases.Methods A total of 1 000 patients who underwent TLIF for lumbar degenerative diseases in our hospital were included and divided into the infection group(n=23)and the non-infection group(n=977)according to whether the surgical incision was infected.General data,surgical and laboratory indicators of patients were collected,and potential risk factors of SSI were screened by univariate analysis and multivariate regression analysis,a nomogram model was established,and its predictive efficiency was validated by the receive operating characteristic(ROC)curve.Results The incidence of SSI in patients after TLIF was 2.3％.The results of univariate analysis showed that age,operative time,intraoperative blood loss,preoperative C-reactive protein(CRP),smoking,and diabetes mellitus were the significant risk factors for the occurrence of SSI.Multivariate regression analysis showed that older age,longer operation time,more intraoperative blood loss,smoking and diabetes mellitus were the independent risk factors for postoperative SSI.ROC curve showed that the nomogram model established in this study has good predictive efficiency.Conclusion Older age,longer operation time,more intraoperative blood loss,smoking,and diabetes mellitus were independent risk factors for postoperative SSI.For patients with these high risk factors,corresponding intervention measures should be taken before operation to reduce the incidence of SSI.

Objective To investigate the population characteristics,distribution of traditional Chinese medicine(TCM)syndromes and influencing factors of long-term prognosis of diarrhea-predominant irritable bowel syndrome(IBS-D),and to provide evidence for the formulation of intervention program for IBS-D patients.Methods A total of 124 patients with IBS-D admitted to the medical institutions of the project team members from July 2020 to August 2022 were selected.According to the scoring results of IBS Quality of Life Measure(IBS-QOL),the patients were divided into the good prognosis group(81 cases)and the poor prognosis group(43 cases).The distribution of TCM syndromes in patients with IBS-D was explored,and the difference of IBS-QOL scores of the patients between good prognosis group and poor prognosis group was compared.Univariate logistic regression analysis and multivariate logistic regression analysis were used to determine the main risk factors for poor prognosis in patients with IBS-D.Results(1)The analysis of population characteristics showed that there was no significant difference in the proportion of male and female patients with IBS-D.The patients with IBS-D were usually middle-aged,and had a large interval span of the course of disease.The severity of their symptoms was mostly moderate.All of the patients with IBS-D had various degrees of anxiety and depression,and had nutritional imbalance.(2)The distribution of TCM syndromes in the patients with IBS-D were shown as the following:78 cases were identified as liver depression and spleen deficiency type,accounting for 62.90％;26 cases were identified as spleen-qi deficiency type,accounting for 20.97％;20 cases were identified as spleen and kidney yang deficiency type,accounting for 16.13％.(3)Analysis of IBS-QOL score showed that compared with the good prognosis group,the items scores of negative emotion,physical function,behavioral disorder,health status,being fastidious about food,social function,sexual behavior and interpersonal relationship of IBS-QOL in the poor prognosis group were significantly lowered(P＜0.01).(4)The univariate analysis showed that the risk of poor prognosis in patients with IBS-D would be increased by the factors of age,education level,course of disease,severity of symptoms,anxiety state,depression state,TCM syndrome types,Acute Physiology and Chronic Health Evaluation scoring system Ⅱ(APACHE 11)score,complication of neurological diseases,hemoglobin level,albumin level and total protein level(P＜0.01).(5)The multivariate Logistic regression analysis showed that the risk factors for poor prognosis of IBS-D patients involved age,education level below junior high school,the severity of symptoms being severe,Self-Rating Anxiety Scale(SAS)score,Self-Rating Depression Scale(SDS)score,TCM syndrome being liver depression and spleen deficiency type,hemoglobin level,albumin level and total protein level(P＜0.01).Conclusion Most of IBS-D patients exert long-term poor prognosis,and their long-term prognosis is affected by the factors of age,education level,severity of symptoms,anxiety and depression state,nutritional imbalance and TCM syndrome being liver depression and spleen deficiency type.The identification of the risk factors of poor prognosis will provide evidence for the formulation and adjustment of clinical intervention programs.