ObjectiveTo develop a machine learning (ML)-based prediction model for neurological rehabilitation outcomes of stroke patients. MethodsA total of 420 stroke patients admitted to the Fuzhou First People's Hospital from October, 2022 to October, 2024 were enrolled as the training set. According to the modified Rankin Scale (mRS) scores three months after discharge, the patients were divided into prognosis group (n = 289) and poor prognosis group (n = 131). An additional 180 stroke patients hospitalized in the same hospital from November, 2024 to April, 2025 were selected as the validation set. Univariate analysis, least absolute shrinkage and selection operator regression, and multivariate logistic regression were performed to identify independent influencing factors for the prognosis of neurological function recovery. Using the screened independent influencing factors as feature variables, six ML models were established, including logistic regression, linear discriminant analysis, naive Bayes, support vector machine, random forest and extreme gradient boosting (XGBoost). The area under the receiver operating characteristic curve (AUC), confusion matrix indicators (accuracy, precision, recall and F1-score), calibration curve and decision curve analysis were adopted to evaluate the predictive efficacy, calibration degree and clinical net benefit of each model, with external validation conducted in the validation set. The SHapley Additive exPlanations framework was used to interpret the optimal model, and bar charts were applied to visualize the feature importance of the best model. ResultsAge, National Institutes of Health Stroke Scale (NIHSS) score, collateral circulation grading, fasting plasma glucose (FPG), lymphocyte percentage (LYMPH%), and homocysteine (Hcy) were independent risk factors for poor neurological rehabilitation prognosis (P < 0.05). For the XGBoost model, the AUC of the training and validation sets were 0.963 (95%CI 0.947 to 0.979) and 0.825 (95%CI 0.764 to 0.885), respectively, while the accuracy was 88.81% and 77.22%, the precision was 92.86% and 68.42%, the recall was 69.47% and 47.27%, and the F1-score was 79.48% and 55.91%, optimal in both calibration and clinical net benefit. The feature importance ranking for the XGBoost model from high to low was NIHSS score, age, collateral circulation grading, FPG, Hcy and LYMPH%. ConclusionThe interpretable XGBoost ML model exhibits excellent predictive efficacy and favorable clinical applicability in predicting neurological rehabilitation outcomes for stroke patients.

In March 2024, three male and two female specimens of the genus Lispe were intercepted on cargo ships inbound for Wan Zai Port under Gongbei Customs. Based on morphological characteristics and molecular analysis, the specimens were identified as Lispe pumila. In this study, we describe the morphological features, diagnostic characteristics, and geographical distribution of L. pumila, with the aim of providing a reference for the identification of this species when intercepted at ports.

Aim To evaluate the bioequivalence of two oral preparations of rivaroxaban tablets(test preparation T and refe-rence preparation R)in fasting/postprandibular state in healthy Chinese subjects.Methods A randomized,open,single-dose,four-cycle,completely repeated crossover experiment was used in this study.A total of 70 healthy male and female subjects were enrolled,including 38 subjects in the fasting group and 32 sub-jects in the postprandial group.Rivaroxaban tablets(2.5 mg/tablet)were taken orally once per cycle and their reference preparations were tested.The plasma rivaroxaban concentration was determined by LC-MS/MS method.The pharmacokinetic parameters of rivaroxaban tablets were calculated by WinNonlin software,and the parameters were analyzed and processed.Re-sults The PK parameters of rivaroxaban tablets and reference preparations in fasting group were as follows:Cmax was(72.48±17.08)and(66.36±15.64)μg·L-1,respectively.AUC0-t were(383.49±101.06)and(370.43±102.16)h·ng·mL-1,and AUC0-inr were(389.58±102.28)and(375.84±103.01)h·μg·L-,respectively.Main PK parameters of subjects taking rivaroxaban tablets orally after meals:Cmax were(66.48±15.64 and 60.87±13.44)μg·L-1,AUC0-t were(404.44±72.58)and(381.80±79.93)h·μg·L-1,re-spectively.AUC0_inf was(410.88±73.55)and(393.64±69.71)h·μg·L-1,respectively.Under fasting and postmeal conditions,subjects took rivaroxaban test and reference prepara-tion orally,one tablet(2.5 mg/tablet)each time.The geometric mean of the main pharmacokinetic parameters of rivaroxaban in plasma(Cmax,AUC0-t,AUC0-inf)and their corresponding values had a 90％confidence interval ranging from 80.00％to 125.00％.No serious adverse events or unexpected adverse e-vents occurred in both groups.Conclusion Rivaroxaban tablets are bioequivalent and safe in vivo under fasting and postprandial conditions.

Objective:To explore the clinical efficacy of the sixth generation CyberKnife (M6) in treating patients with brain metastases, and analyze clinical characteristics and dosimetric factors.Methods:Clinical data of patients with brain metastases who received CyberKnife treatment at Sichuan Cancer Hospital from April 2023 to March 2024 were retrospectively analyzed. All patients were treated with CyberKnife with 6 MV X-ray. According to the maximum diameter of brain metastases, the radiation prescription dose of brain metastases was adjusted. The tumor remission, recurrence, 6-month and 1-year overall survival (OS), local control (LC) of intracranial target lesions, progression-free survival (PFS), distant metastasis-free survival (DMFS) of intracranial brain metastases and adverse reactions were evaluated. According to the median biological dose, the survival difference between the groups was compared. Survival analysis was conducted by Kaplan-Meier method. Survival differences among different groups were analyzed by log-rank test.Results:A total of 63 eligible patients with brain metastases were enrolled, with a median age of 59 years (rang: 36-80 years). Among them, 47 patients were diagnosed with primary tumors originating from the lungs, 16 patients with primary tumors originating from other organs; 44 patients with single brain metastases, and 19 patients with 2-3 lesions, respectively. The median biological dose was 67.2 Gy (rang: 47.4-86.4 Gy), and the median single dose was 8 Gy/F (rang: 4-24 Gy/F). The follow-up was conducted until July 15, 2024. The median follow-up time for the entire group was 9 months (rang: 2-15 months). Among the 87 target lesions treated with CyberKnife, 11 patients corresponding to 14 target lesions experienced local recurrence. And the 6-month and 1-year LC rates were 92.5% and 70.9%, respectively. Ten patients corresponding to 16 target lesions died. And the 6-month and 1-year OS rates were 92.7% and 74.8%, respectively. Thirty-five patients corresponding to 50 target lesions experienced disease progression. And the 6-month and 1-year PFS rates were 64.3% and 25.5%, respectively. Thirty-three patients corresponding to 48 target lesions showed distant metastasis outside the target lesions, with a 6-month DMFS of 67.0% and a 1-year DMFS of 33.9%. Group comparison showed that 43 target lesions in the group receiving ≤67.2 Gy irradiation and 44 in the group receiving >67.2 Gy irradiation. The 6-month LC, OS, PFS, and DMFS rates between two groups were 89.8% vs. 97.7% ( P=0.127), 89.8% vs. 95.4% ( P=0.305), 65.4% vs. 68.5% ( P=0.514), 65.4% vs. 68.5% ( P=0.516), respectively. The 1-year LC, OS, PFS, and DMFS rates between two groups were 54.1% vs. 89.5% ( P=0.003), 67.3% vs. 82.9% ( P=0.219), 19.2% vs. 32.7% ( P=0.370) and 23.3% vs. 33.0% ( P=0.533). During the follow-up, only 2 patients (3.2%) were found to have grade 1-2 radiation-induced brain injury (asymptomatic brain injury) by MRI examination, and there were no other radiotherapy related adverse reactions. Conclusions:CyberKnife therapy is clinically effective for brain metastases, with mild adverse reactions. Increasing the tumor irradiation dose can improve local tumor control and is expected to further improve the OS of patients.

Objective:To analyze the influencing factors of pregnancy outcome of test-tube baby assisted pregnancy base on clinical data,and build a prediction model.Methods:The clinical data of 283 infertility patients were retrospectively analyzed.They were divided into successful pregnancy group(147 cases)and failed pregnancy group(136 cases)according to the pregnancy outcome.The influencing factors of pregnancy were analyzed by multivariate Logistic regression analysis.The predictive value of influencing factors was evaluated by drawn receiver operating characteristic(ROC)curve.Prediction model was constructed and verified based on the results of multivariate Logistic regression analysis.Results:283 cases of infertility patients received test-tube baby assisted pregnancy treatment,147 cases were successfully pregnant,and the pregnancy success rate was 51.94％(147/283).The proportion of the duration of infertility＞3 years,the proportion of endometrial morphology on HCG injection day was B or C,the proportion of endometrial thickness＜7 mm,and the proportion of embryo quality not good in failed pregnancy group were higher than those in successful pregnancy group(P＜0.05).Endometrial thickness＜7 mm,endometrial morphology on HCG injection day was B or C,the duration of infertility＞3 years,embryo quality not good were the risk factors for the failure of test-tube baby assisted pregnancy(P＜0.05).The results of ROC curve analysis showed that,the area under the curve(AUC)of endometrial morphology on HCG injection day,the duration of infertility,endometrial thickness and embryo quality to predict the failure of test-tube baby assisted pregnancy were 0.577,0.617,0.581 and 0.603 respectively.Multivariate Logistic regression model was constructed based on the influencing factors of failure of test-tube baby assisted pregnancy,the AUC of this model for predicting failure of test-tube baby assisted pregnancy was 0.845,and its predictive efficacy was higher than that of each influencing factor alone(Z=6.832,5.905,6.832,6.258,all P＜0.05).Conclusion:Endometrial thicknes＜7 mm,endometrial morphology on HCG injection day was B or C,the duration of infertility＞3 years,embryo quality not good are risk factors for the failure of test-tube baby assisted pregnancy,the prediction model base on the above risk factors has a high predictive value for the pregnancy outcome of test-tube baby assisted pregnancy.

Aim To evaluate the bioequivalence of two oral preparations of rivaroxaban tablets(test preparation T and refe-rence preparation R)in fasting/postprandibular state in healthy Chinese subjects.Methods A randomized,open,single-dose,four-cycle,completely repeated crossover experiment was used in this study.A total of 70 healthy male and female subjects were enrolled,including 38 subjects in the fasting group and 32 sub-jects in the postprandial group.Rivaroxaban tablets(2.5 mg/tablet)were taken orally once per cycle and their reference preparations were tested.The plasma rivaroxaban concentration was determined by LC-MS/MS method.The pharmacokinetic parameters of rivaroxaban tablets were calculated by WinNonlin software,and the parameters were analyzed and processed.Re-sults The PK parameters of rivaroxaban tablets and reference preparations in fasting group were as follows:Cmax was(72.48±17.08)and(66.36±15.64)μg·L-1,respectively.AUC0-t were(383.49±101.06)and(370.43±102.16)h·ng·mL-1,and AUC0-inr were(389.58±102.28)and(375.84±103.01)h·μg·L-,respectively.Main PK parameters of subjects taking rivaroxaban tablets orally after meals:Cmax were(66.48±15.64 and 60.87±13.44)μg·L-1,AUC0-t were(404.44±72.58)and(381.80±79.93)h·μg·L-1,re-spectively.AUC0_inf was(410.88±73.55)and(393.64±69.71)h·μg·L-1,respectively.Under fasting and postmeal conditions,subjects took rivaroxaban test and reference prepara-tion orally,one tablet(2.5 mg/tablet)each time.The geometric mean of the main pharmacokinetic parameters of rivaroxaban in plasma(Cmax,AUC0-t,AUC0-inf)and their corresponding values had a 90％confidence interval ranging from 80.00％to 125.00％.No serious adverse events or unexpected adverse e-vents occurred in both groups.Conclusion Rivaroxaban tablets are bioequivalent and safe in vivo under fasting and postprandial conditions.

BACKGROUND:In recent years,artificial intelligence has gradually emerged and has been applied in various fields such as spinal cord nerve injury and repair,which has a positive impact on clinical treatment. OBJECTIVE:To study the application progress of artificial intelligence in the diagnosis,treatment,and rehabilitation of spinal cord nerve injury and repair,clarify the research hotspots and shortcomings in this field,and provide suggestions for future research work. METHODS:Relevant literature on artificial intelligence in the field of spinal cord nerve injury and repair was retrieved on the Web of Science core collection database until 2023.CiteSpace 6.1.R6 and VOSviewer 1.6.19 software was used to perform general literature analysis,co-citation of literature,co-citation of journals,double image overlay of journals,keyword clustering,and other visual analysis on the literature data. RESULTS AND CONCLUSION:(1)A total of 1 713 articles were selected,and the annual publication volume in this field showed a fluctuating upward trend,with the United States taking the lead,and Kadone and Hideki being the authors with the highest publication volume.ARCH PHYS MED REHAB was the journal with the highest number of citations.(2)Keyword co-occurrence and cluster analysis showed that after removing keywords similar to the search terms,the main keywords were divided into three main clusters:Exoskeleton and exercise rehabilitation(the largest core hotspot);machine learning and neural plasticity;robotics and rehabilitation training.(3)Keyword burst analysis showed that deep learning and artificial intelligence had become burst terms in the past five years.(4)The results of in-depth analysis of co cited and highly cited literature showed that the hotspots of artificial intelligence in the field of spinal cord nerve injury and repair were mainly focused on powered exoskeletons,gaits,electrical nerve stimulation,intracortical brain-computer interface(IBCI),robots,and polymer biomaterials,and neural stem cell.(5)The research on artificial intelligence in the field of spinal cord nerve injury and repair has shown an upward trend in recent years.The focus of this field had gradually shifted from single treatment methods such as exoskeletons and electrical stimulation to intelligent,precise,and personalized directions.(6)There were some limitations in this field,such as the consequences of missing or imbalanced data,low data accuracy and reproducibility,and ethical issues(such as privacy,research transparency,and clinical reliability).Future research should address the issue of data collection,requiring large sample,high-quality clinical datasets to establish effective artificial intelligence models.At the same time,the research on genomics and other mechanisms in this field is very weak.In the future,various machine learning technologies such as brain chips can be used,and gene editing therapy,single-cell spatial transcriptome and other methods can be used to study the basic mechanisms of regeneration-related gene upregulation and axon growth structural protein production.

Objective:To investigate the influence of inner cell mass and trophectoderm cytoplasmic strings during blastocyst expan-sion on embryonic development and pregnancy outcome.Methods:A retrospective cohort analysis was performed for the clinical data of patients who received pre-implantation genetic testing for aneuploidy(PGT-A)and underwent single blastocyst transplantation in our hospital from June 2019 to December 2021.A total of 530 patients were enrolled,and genetic testing was performed for 2132 blasto-cysts.According to the presence or absence of cytoplasmic strings during blastocyst expansion,the blastocysts were divided into cyto-plasmic strings(+)group with 534 blastocysts and cytoplasmic strings(-)group with 1598 blastocysts,and quality and PGT-A results were compared between the two groups.After the transfer of euploid blastocysts,pregnancy outcome was compared between the 115 blastocysts with cytoplasmic strings and the 415 blastocysts without cytoplasmic strings.Results:The rates of cytoplasmic strings(+)in the high-,average-,and low-quality blastocyst groups were 30.19%,24.62%,and 12.63%,respectively.The correlation analysis showed a correlation coefficient of-0.115(P<0.001)between embryo quality and the rate of cytoplasmic strings(+).There was no sig-nificant difference in euploidy rate between the two groups(45.3%vs.44.6%).There were no significant differences between the euploid blastocysts with cytoplasmic strings and those without cytoplasmic strings in implantation rate(72.17%vs.66.02%,P=0.213),miscarriage rate(14.46%vs.12.77%,P=0.691),and live birth rate(61.74%vs.57.59%,P=0.424).Conclusion:The presence of cyto-plasmic strings is associated with the morphological quality of blas-tocysts,while it has no impact on embryo ploidy or clinical outcome after euploid embryo transfer.Further research is needed to confirm the impact of cytoplasmic strings on embryonic development.

Objective:To explore the technique methods and clinical application value of β-electrode (a plasma needle shape electrode) assisted laparoscopic repair of complex vesicovaginal fistula (VVF).Methods:Clinical data of one patient with complex VVF admitted to Chinese PLA General Hospital in April 2025 was retrospectively analyzed. A 36-year-old female presented with urinary leakage 2 months after hysterectomy. Computed tomography urography excluded ureterovaginal fistula. Cystoscopy revealed a 2 cm fistula on the posterior bladder wall with both ureteral orifices adjacent to the fistula edge. The procedure involved two steps: first, transurethral β-electrode pretreatment was performed to protect the ureteral orifices and create a passage from the bladder to the abdominal cavity. Then, laparoscopic separation, suture closure of the fistula, and omental flap coverage were conducted.Results:Total operation time was 180 min (the time of β- electrode operation was 30 min) with intraoperative estimated blood loss of 50 mL. The catheter was removed 3 weeks postoperatively, and the patient voided well without leakage during 4-month follow-up.Conclusions:β- electrode assisted laparoscopic repair of complex VVF have the advantage of precise manipulation, minimal invasion and rapid recovery. No similar technique have been reported domestically or internationally. This technique provides a new approach for the treatment of complex VVF.

Objective To explore the effectiveness and safety of hypofractionated radiotherapy(HFRT)combined with transcatheter arterial chemoembolization(TACE)for metachronous hepatic oligometastasis from nasopharyngeal carcinoma(NPC).Methods The clinical data of patients with metachronous hepatic oligometastasis from NPC,who received HFRT combined with TACE treatment at the Sichuan Provincial Cancer Hospital of China from January 2012 to October 2022,were retrospectively analyzed.The 3-year and 5-year overall survival(OS),progression-free survival(PFS),local recurrence-free survival(LRFS),no extrahepatic distant metastasis survival(EMFS),and treatment-related adverse reactions were analyzed.Results A total of 55 patients were enrolled in this study,including 36 males and 19 females,the median age at the time of occurring metachronous hepatic oligometastasis was 44(27-75)years.The 3-year and 5-year OS,PFS,LRFS,EMFS were 67.8％and 40.0％,55.8％and 30.0％,72.7％and 56.5％,63.6％and 56.5％,respectively.The subgroup analysis indicated that the treatment course of TACE ≥3 cycles could significantly improve the PFS of patients with oligometastasis.HFRT combined with TACE treatment was well tolerated by all patients,and the incidence of Grade Ⅲ-Ⅳadverse reactions was quite low.Conclusion For the treatment of metachronous hepatic oligometastasis from NPC,HFRT combined with TACE is clinically effective,besides,the patients can well tolerate the therapeutic scheme.