The risk prediction of paroxysmal atrial fibrillation (PAF) is a challenge in the field of biomedical engineering. This study integrated the advantages of machine learning feature engineering and end-to-end modeling of deep learning to propose a PAF risk prediction method based on multimodal feature fusion. Additionally, the study utilized four different feature selection methods and Pearson correlation analysis to determine the optimal multimodal feature set, and employed random forest for PAF risk assessment. The proposed method achieved accuracy of (92.3 ± 2.1)% and F1 score of (91.6 ± 2.9)% in a public dataset. In a clinical dataset, it achieved accuracy of (91.4 ± 2.0)% and F1 score of (90.8 ± 2.4)%. The method demonstrates generalization across multi-center datasets and holds promising clinical application prospects.
Humans ; Atrial Fibrillation/diagnosis* ; Machine Learning ; Deep Learning ; Risk Assessment/methods*

OBJECTIVES: To create a mixed valvular heart disease (MVHD)-related age-adjusted comorbidity index (MVACI) model for predicting mortality risk of patients with MVHD. METHODS: A total of 4080 patients with moderate or severe MVHD in the China-VHD study were included. The primary endpoint was 2-year all-cause mortality. A MVACI model prediction model was constructed based on the mortality risk factors identified by univariate and multivariate Cox regression analysis. Restricted cubic splines were used to assess the relationship between MVACI scores and 2-year all-cause mortality. The optimal threshold, determined by the maximum Youden index from receiver operator characteristic (ROC) curve analysis, was used to stratify patients. Kaplan-Meier method was used to calculate 2-year all-cause mortality and compared using the Log-rank test. Univariate and multivariate Cox proportional hazards models were employed to calculate hazard ratios (HR) and 95% confidence intervals (CI), evaluating the association between MVACI scores and mortality. Paired ROC curves were used to compare the discriminative ability of MVACI scores with the European System for Cardiac Operative Risk Evaluation Ⅱ(EuroSCORE Ⅱ) or the age-adjusted Charlson comorbidity index (ACCI) in predicting 2-year clinical outcomes, while calibration curves assessed the calibration of these models. Internal validation was performed using the Bootstrap method. Subgroup analyses were conducted based on etiology, treatment strategies, and disease severity. RESULTS: Multivariate analysis identified the following variables independently associated with 2-year all-cause mortality in patients: pulmonary hypertension, myocardiopathy, heart failure, low body weight (body mass index <18.5 kg/m²), anaemia, hypoalbuminemia, renal insufficiency, cancer, New York Heart Association (NYHA) class and age. The score was independently associated with the risk of all-cause mortality, and exhibited good discrimination (AUC=0.777, 95%CI: 0.755-0.799) and calibration (Brier score 0.062), with significantly better predictive performance than EuroSCORE Ⅱ or ACCI (both adjusted P<0.01). The internal validation showed that the MVACI model's predicted probability of 2-year all-cause mortality was generally consistent with the actual probability. The AUCs for predicting all-cause mortality risk were all above 0.750, and those for predicting adverse events were all above 0.630. The prognostic value of the score remained consistent in patients regardless of their etiology, therapeutic option, and disease severity. CONCLUSIONS The MVACI was constructed in this study based on age and comorbidities, and can be used for mortality risk prediction and risk stratification of MVHD patients. It is a simple algorithmic index and easy to use.
Humans ; Prognosis ; Comorbidity ; Heart Valve Diseases/epidemiology* ; Female ; Male ; Middle Aged ; Aged ; Proportional Hazards Models ; Risk Factors ; China/epidemiology* ; Age Factors ; Risk Assessment ; Adult ; ROC Curve

Objective:To investigate the efficacy and safety of lenvatinib plus drug-eluting bead transarterial chemoembolization (DEB-TACE) and FOLFOX-based hepatic arterial infusion chemotherapy (Len+DEB-TACE+HAIC) versus lenvatinib plus DEB-TACE (Len+DEB-TACE) for hepatocellular carcinoma (HCC) larger than 7 cm with portal vein tumor thrombosis (PVTT).Methods:The data from patients diagnosed with HCC (>7 cm) and PVTT who received either Len+DEB-TACE+HAIC ( n=99) or Len+DEB-TACE ( n=102) between July 2019 and June 2021 at six institutions in China were collected and retrospectively analyzed. Tumor responses were evaluated based on modified Response Evaluation Criteria in Solid Tumors. Objective response rate (ORR), disease control rate (DCR), time to progression (TTP), overall survival (OS), and treatment-related adverse event (TRAE) were compared between the two groups by propensity score matching. Subgroup analyses were performed for TTP and OS. Results:After propensity score matching, 83 pairs of patients were included in the study cohorts. The ORR for the Len+DEB-TACE+HAIC group and the Len+DEB-TACE group was 66.3% and 38.6% ( χ2=12.78, P<0.001), respectively. The DCR for the Len+DEB-TACE+HAIC group and the Len+DEB-TACE group was 91.6% and 79.5% ( χ2=4.87, P=0.027), respectively. The median TTP and median OS for the Len+DEB-TACE+HAIC group were significantly longer than those for the Len+DEB-TACE group (TTP, 10.1 months vs. 6.1 months, χ2=35.28, P<0.001; OS, 17.3 months vs. 12.9 months, χ2=16.84, P<0.001). The incidence of ≥grade 3 TRAEs was 38.6% in the Len+DEB-TACE+HAIC group and 33.7% in the Len+DEB-TACE group ( χ2=0.42, P=0.518). Conclusion:Compared with Len+DEB-TACE, Len+DEB-TACE+HAIC led to improved tumor response, TTP and OS with an acceptable safety profile in patients with large HCC and PVTT.

A retrospective investigation was conducted to analyze the occupational exposure history, clinical manifestations, laboratory tests, imaging findings, and diagnosis and treatment of two cases of sequential latent occupational acute organotin poisoning. Both patients were successively employed in the same enterprise, engaged in crushing of waste polyvinyl chloride plastics, and thus potentially exposed to organotin hazards. Within several days of employment, both patients developed discomfort symptoms, and central nervous system impairment was observed, including short-term memory loss, slow response, and cognitive dysfunction. Hypokalemia was detected in both cases. Cranial magnetic resonance imaging showed abnormalities (multiple ischemic lesions in the bilateral frontal and parietal lobes), and urinary tin was positive. Symptoms relieved in both patients after treatments with tin-exclusion, potassium supplementation, and neurotrophic treatment. Based on the GBZ 26-2007 Diagnostic Criteria of Occupational Acute Trialkyltin Poisoning, and combined with worksite survey of occupational health and exclusion of cerebrovascular disease, viral encephalitis, and autoimmune encephalitis and other neurological disorders, both patients were diagnosed with mild occupational acute trialkyltin poisoning. Sequential latent occupational acute organotin poisoning is prone to misdiagnosis, with great difficulty in etiological identification. Comprehensive assessment of occupational exposure history and biomarker testing are essential for differential diagnosis. Early recognition and intervention improve prognosis, highlighting the need for strengthened occupational health supervision and protection in high-risk work posts.

This paper presents an automatic acquisition and analytic procedure of acupuncture manipulation based on optical navigation, aiming at solving the shortcomings of existing acquisition methods of acupuncture manipulation. An acquisition holder installed at the handle tail of filiform needle was designed to display the movement trajectory of the needle during acupuncture delivery by collecting the movement trajectory of holder. The 3-month old male Bama miniature pig was selected as the experimental subject, and 6 points, "Bojian" "Qiangfeng" "Housanli" "Xiaokua" "Huiyang" (BL35) and "Baihui" (GV20), were selected during acupuncture manipulation. The optical navigation system was used to collect the real-time data, and these data were per-processed and analyzed using mean filtering and Fourier transform. The acupuncture procedure was divided into 3 stages, inserting, lifting-thrusting, and twisting. The results showed that the accuracy was 96.3% at lifting-thrusting stage, and that was 100.0% at twisting stage. The decomposition effect of the entire procedure was satisfactory. This study provides a new approach to the quantitative analysis of acupuncture manipulation. In the future, it needs to further optimize the algorithm and expand the sample size so as to improve the accuracy of this analytic technique.
Acupuncture Therapy/methods* ; Male ; Animals ; Swine ; Acupuncture Points ; Humans ; Swine, Miniature ; Needles

BACKGROUND: Valvular heart disease (VHD) has become increasingly common with the aging in China. This study aimed to evaluate regional differences in the clinical features, management strategies, and outcomes of patients with VHD across different regions in China. METHODS: Data were collected from the China-VHD Study. From April 2018 to June 2018, 12,347 patients who presented with moderate or severe native VHD with a median of 2 years of follow-up from 46 centers at certified tertiary hospitals across 31 provinces, autonomous regions, and municipalities in Chinese mainland were included in this study. According to the locations of the research centers, patients were divided into five regional groups: eastern, southern, western, northern, and central China. The clinical features of VHD patients were compared among the five geographical regions. The primary outcome was all-cause mortality or rehospitalization for heart failure. Kaplan-Meier survival analysis was used to compare the cumulative incidence rate. RESULTS: Among the enrolled patients (mean age, 61.96 years; 6877 [55.70%] male), multiple VHD was the most frequent type (4042, 32.74%), which was mainly found in eastern China, followed by isolated mitral regurgitation (3044, 24.65%), which was mainly found in northern China. The etiology of VHD varied significantly across different regions of China. The overall rate of valve interventions was 32.67% (4008/12,268), with the highest rate in southern China at 48.46% (205/423). In terms of procedure, the proportion of transcatheter valve intervention was relatively low compared to that of surgical treatment. Patients with VHD in western China had the highest incidence of all-cause mortality or rehospitalization for heart failure. Valve intervention significantly improved the outcome of patients with VHD in all five regions (all P  <0.05). CONCLUSIONS: This study revealed that patients with VHD in China are characterized by significant geographic disparities in clinical features, treatment, and clinical outcomes. Targeted efforts are needed to improve the management and prognosis of patients with VHD in China according to differences in geographical characteristics. REGISTRATION ClinicalTrials.gov , NCT03484806.
Aged ; Female ; Humans ; Male ; Middle Aged ; China/epidemiology* ; Heart Valve Diseases/therapy* ; Kaplan-Meier Estimate ; Tertiary Care Centers ; Treatment Outcome

Objective:To investigate the correlation between non-Hodgkin lymphoma (NHL) and chronic hepatitis B virus (HBV) infection by using the method of two-sample bidirectional Mendelian randomization (MR) analysis.Methods:Genetic variation data for NHL came from the Finnish database (FinnGen) Consortium 2021 public genome-wide association study (GWAS) dataset including 1 088 patients with NHL and 299 952 control subjects. The GWAS dataset for chronic HBV infection was derived from GWAS analysis published in 2021, including 145 NHL patients and 351 740 control subjects. NHL was used as an exposure factor, single nucleotide polymorphism (SNP) significantly associated with NHL was used as an instrumental variable (IV), chronic HBV infection was used as an outcome variable. The two-sample MR analysis was performed by using inverse-variance weighted (IVW) method. Chronic HBV infection was taken as an exposure factor, SNP significantly associated with chronic HBV infection was taken as IV, and NHL was taken as outcome variable, and then reverse two-sample MR analysis was performed. The IVW method used the inverse variance of each IV as the weight to fit, and the ratio method was used to measure SNP one by one and make weighted regression analysis, so as to obtain the overall estimate. MR-Egger regression and the weighted median (WME) method were also used to supplement the IVW method. In sensitivity analysis, leave-one-out sensitivity analysis was used to evaluate the impact of a single SNP. Cochran Q test was used to analyze the heterogeneity of the selected IV. MR-Egger regression was used to measure the average horizontal pleiotropy of IV, and the P-value of directivity was calculated. The MR-pleiotropy residual sum and outlier (MR-PRESSO) Global Test was used to exclude possible horizontal pleiotropic outliers and reduce bias. Results:In the leave-one-out sensitivity analysis, SNP with significant effects on causal associations was excluded. In forward MR analysis, IVs were 10 SNPs associated with NHL; the IVW method indicated that there was no causal association between NHL and chronic HBV infection ( OR = 0.979, 95% CI: 0.925-1.036, P = 0.465). MR-Egger regression ( OR = 0.992, 95% CI: 0.926-1.062, P = 0.825) and WME method ( OR = 0.992, 95% CI: 0.934-1.055, P = 0.805) were used as supplementary methods to obtain the consistent results. In sensitivity analysis, Cochran Q test showed no heterogeneity among IVs (IVW method: P = 0.271, MR-Egger regression: P = 0.239). Horizontal pleiotropy was not found in the MR-Egger regression (intercept was -0.01, P = 0.778) and the MR-PRESSO Global Test ( P > 0.05), suggesting robust results. In the reverse MR analysis, IVs were 8 SNPs associated with NHL; the IVW method ( OR = 1.117, 95% CI: 0.942-1.324, P = 0.202) also found no significant causal relationship between chronic HBV infection and NHL; MR-Egger regression ( OR = 0.777, 95% CI: 0.450-1.343, P = 0.401) and WME method ( OR = 1.120, 95% CI: 0.887-1.415, P = 0.351) also showed similar risk estimates. Sensitivity analysis also suggested the consistency and reliability of the results. Cochran Q test showed no heterogeneity among IVs (IVW method: P = 0.775, MR-Egger regression: P = 0.903). Horizontal pleiotropy was not found by MR-Egger regression (intercept was 0.102, P = 0.548) and MR-PRESSO Global Test ( P > 0.05). Conclusions:MR analysis suggests no causal relationship between NHL and chronic HBV infection.

Objective:To develop an artificial intelligence (AI)-based platelet review strategy to identify abnormal platelet histograms with no significant difference between initial impedance platelet count (PLT-I) and PLT-F results.Methods:This study included 5 119 routine blood analysis in Nanfang Hospital of Southern Medical University and its Ganzhou branch from July 2023 and March 2024. Specimens exhibiting abnormal platelet histograms and an initial platelet count >40×10?/L underwent review using the fluorescent platelet count (PLT-F) channel. Consistency of the results was defined as a difference between impedance platelet count (PLT-I) and PLT-F less than ±20% of the PLT-F results. A deep learning model was developed using platelet and red blood cell histogram data from a training set of 3 807 specimens. The model′s diagnostic performance was evaluated on an independent external validation set ( n=805) using receiver operating characteristic (ROC) curve analysis. Changes in the number of reviewed samples and sample turnaround time were analyzed to assess its clinical utility. Results:The deep learning model based on platelet and red blood cell histograms achieved an area under the ROC curve (AUC) of 0.854 in the training set. At a cutoff value of 0.1, the sensitivity was 0.954 and specificity was 0.358. The model could reduce review by 16.80% (190/1 131). In the validation set, the AUC was 0.805, with a sensitivity of 0.955 and specificity of 0.307, corresponding to a reduction of 17.41% (47/270) in reviewed specimens.Conclusion:The platelet review prediction model developed based on deep learning technology can efficiently identify samples with consistent results before and after review, reducing unnecessary reviews and shortening specimen testing time, thereby improving the efficiency of platelet test.

Objective:To explore the advantages of robotic-arm assisted total hip arthroplasty (rTHA) in acetabular component positioning and lower limb length assessment in patients with severe pelvic coronal tilt.Methods:A retrospective analysis was conducted on the data of 122 patients with unilateral end-stage hip disease and coronal pelvic tilt angle >3° who underwent total hip arthroplasty (THA) at the First Medical Center of PLA General Hospital from June 2022 to December 2023. Among them, 44 patients underwent rTHA, and 78 underwent manual THA (mTHA). The rTHA group included 18 males and 26 females, with an average age of 60.5±9.3 years; the mTHA group included 41 males and 37 females, with an average age of 58.5±8.4 years. Compare the differences in the anteversion angle, abduction angle, pelvic tilt angle, leg length discrepancy (LLD) of the acetabular prosthesis, and the proportions of patients with LLD>0.5 cm and >1 cm between the two groups of patients after surgery. Calculate the proportion of outlier rates of acetabular abduction angle (<30° or >45°), and proportions within Callanan's safe zone. The early efficacy of the hip joint was evaluated by using the modified Harris score and joint range of motion.Results:All patients were followed up for 6 to 12 months, with an average of 8 months. All the surgical incisions of the patients achieved primary healing. Postoperative comparisons showed no statistically significant differences in acetabular abduction angle (39.5°±3.3° vs. 38.4°±7.3°) or anteversion angle (20.7°±1.6° vs. 19.7°±1.6°) between rTHA and mTHA groups ( P>0.05). However, pelvic tilt angle [2.5° (1.1°, 3.6°) vs. 3.5° (2.3°, 5.9°)] showed a statistically significant difference ( U=4.371, P=0.008). The rTHA group exhibited smaller absolute LLD [0.2 (0.1, 0.4) cm vs. 0.5 (0.2, 0.5) cm] and lower proportions of LLD >0.5 cm [14% (6/44) vs. 49% (38/78)] and >1 cm [2% (1/44) vs. 18% (14/78)], with statistical significance ( P<0.05). The rTHA group had a lower outlier rate for acetabular abduction angle (<30°or >45°) compared to the mTHA group [2% (1/44) vs. 33% (26/78)], with statistical significance (χ 2=10.388, P<0.001). Taking the Callanan safety zone as the standard, the proportion of acetabular cups within the safe zone was significantly higher in the rTHA group (98%, 43/44) compared to the mTHA group (67%, 52/78), with a statistically significant difference (χ 2=13.998, P<0.001). The modified Harris score and hip joint range of motion in the mTHA group increased from 47.6±6.6 points and 83° (73°, 88°) before the operation to 83.5±11.2 points and 118° (110°, 122°) at the last follow-up, respectively. The rTHA group increased from 46.5±9.2 points and 79° (71°, 90°) before the operation to 85.0±12.5 points and 124° (116°, 130°) at the last follow-up. The differences in the modified Harris score and hip joint range of motion between the two groups before the operation and at the last follow-up were statistically significant ( P<0.05). However, there was no statistically significant difference between the groups at the last follow-up ( P>0.05). No THA-related complications occurred during follow-up period. Conclusion:For patients with end-stage hip joint diseases with coronal tilt exceeding 3°, robotic-assisted technology significantly improves the accuracy of acetabular component placement during THA and offers better control of postoperative LLD.

Objective:To construct an infectious clone of a Chinese HIV-1 CRF01_AE epidemic strain with strong replication capacity，and comprehensively identify its viral phenotype and replication capacity.Methods:Using the CRF01_AE clinical isolate GX2005002，which was previously isolated from whole blood of an HIV-1-infected individual in China by our laboratory，as the parental strain，the full-length genome of the virus（9.7 kb）was divided into 5' half fragment（5.1 kb）and 3' half fragment（4.6 kb）for amplification. The proviral DNA was used as a template to amplify the virus genome，which was then ligated into the eukaryotic expression vector pEASY-T1. The consistency of its sequence with the parental strain sequence was identified through full-length genome sequencing and phylogenetic analysis. The replication capacity，syncytium inducibility，and cell tropism were experimentally identified to determine its phenotypic consistency with the parental strain.Results:An infectious clone of the CRF01_AE strain was successfully constructed，and its genome sequence exhibited high consistency with the sequence of the parental strain. By transfecting target cells，a derivative virus with infectious activity and replication capability was successfully rescued. The derived virus maintained phenotypic characteristics consistent with the parental strain，such as cell tropism and syncytium inducibility.Conclusion:This study successfully constructed an infectious clone of a Chinese HIV-1 CRF01_AE epidemic strain with clear background and distinct phenotype. The genomic sequence and viral phenotypic characteristics of the derived virus are basically consistent with the parental strain，providing strong representation of the original isolate and serving as a powerful tool for research on the correlation between the genetic characteristics，viral phenotype，and pathogenicity of HIV-1 CRF01_AE strains.