Cardiotocography (CTG) is a non-invasive and important tool for diagnosing fetal distress during pregnancy. To meet the needs of intelligent fetal heart monitoring based on deep learning, this paper proposes a TWD-MOAL deep active learning algorithm based on the three-way decision (TWD) theory and multi-objective optimization Active Learning (MOAL). During the training process of a convolutional neural network (CNN) classification model, the algorithm incorporates the TWD theory to select high-confidence samples as pseudo-labeled samples in a fine-grained batch processing mode, meanwhile low-confidence samples annotated by obstetrics experts were also considered. The TWD-MOAL algorithm proposed in this paper was validated on a dataset of 16 355 prenatal CTG records collected by our group. Experimental results showed that the algorithm proposed in this paper achieved an accuracy of 80.63% using only 40% of the labeled samples, and in terms of various indicators, it performed better than the existing active learning algorithms under other frameworks. The study has shown that the intelligent fetal heart monitoring model based on TWD-MOAL proposed in this paper is reasonable and feasible. The algorithm significantly reduces the time and cost of labeling by obstetric experts and effectively solves the problem of data imbalance in CTG signal data in clinic, which is of great significance for assisting obstetrician in interpretations CTG signals and realizing intelligence fetal monitoring.
Humans ; Pregnancy ; Female ; Cardiotocography/methods* ; Deep Learning ; Neural Networks, Computer ; Algorithms ; Fetal Monitoring/methods* ; Heart Rate, Fetal ; Fetal Distress/diagnosis* ; Fetal Heart/physiology*

Objective:To analyze the diagnostic and prognostic value of routine bone marrow examination in patients with extranodal NK/T-cell lymphoma (ENKTCL) based on PET-CT staging.Methods:Clinical data of 186 patients who received bone marrow biopsy and bone marrow aspiration in Fujian Medical University Union Hospital from 2013 to 2021 were retrospectively analyzed. All patients were divided into bone marrow biopsy + bone marrow aspiration group ( n=186) and PET-CT + bone marrow biopsy group ( n=139). The sensitivity, specificity, positive and negative predictive values were compared between two groups. The data were analyzed and plotted. Survival analysis was performed using Kaplan-Meier method and log-rank test. Results:In the whole cohort, 45 patients were positive for bone marrow biopsy, and 30 of them were positive for bone marrow aspiration. A total of 141 patients who were negative for bone marrow biopsy also achieved negative results for bone marrow aspiration. A total of 139 patients completed PET-CT staging and bone marrow biopsy. And 30 patients were diagnosed with positive bone marrow by PET-CT, in which 22 of them were confirmed positive by bone marrow biopsy. Among 109 patients diagnosed with negative bone marrow by PET-CT, 5 of them were confirmed positive by bone marrow biopsy. All these cases were classified as stage Ⅳ due to distant metastases. PET-CT had a diagnostic sensitivity of 81.5%, a specificity of 92.9%, a positive predictive value of 73.3%, and a negative predictive value of 95.4%. Among early stage (Ⅰ-Ⅱ stage) patients diagnosed with PET-CT, all of them were negative for bone marrow biopsy (the negative predictive value was 100%). In stage Ⅳ patients ( n=55), the 1-year overall survival of patients with bone marrow involvement by bone marrow biopsy or PET-CT ( n=35) compared with their counterparts with the involvement of other organs ( n=20) was 28.7% vs.42.0% ( P=0.13), and 1-year progression free survival rates was 23.2% vs. 23.3% in ( P=0.94). Conclusions:Routine bone marrow biopsy does not change the original staging of patients with early stage ENKTCL based on PET-CT staging. Advanced stage patients with positive bone marrow biopsy tend to obtain worse prognosis, indicating that bone marrow biopsy still has certain value.

Objective:To explore the effect of continuous blood purification (CBP) on the immunity and endothelial cell function of patients with sepsis.Methods:A prospective study was conducted. The patients aged ≥18 years old and meeting the diagnostic criteria of sepsis admitted to the department of critical care medicine of Binzhou Medical University Hospital from March 2019 to October 2020 were selected as the research subjects, and the patients were divided into standard treatment group and CBP treatment group according to random number table method. Both groups were given standard treatment including initial fluid resuscitation, infection source control and antibiotics according to the 2016 international guidelines for the management of sepsis and septic shock. CBP treatment group was additionally given continuous veno-venous hemofiltration (CVVH) at a dose of 25-30 mL·kg -1·h -1 and blood flow rate of 150-200 mL/min for more than 20 hours a day for 3 days. The clinical data of patients including blood lactic acid (Lac), procalcitonin (PCT), lymphocyte count (LYM), acute physiology and chronic health evaluationⅡ(APACHEⅡ) score, sequential organ failure assessment (SOFA) score were recorded before treatment and 1 day and 3 days after treatment. At the same time, the venous blood was collected, and the immune function related indexes [interleukins (IL-4, IL-7), programmed death receptor-1 (PD-1), programmed death ligand-1 (PD-L1), interferon-γ (IFN-γ)] and endothelial cell injury related markers [soluble thrombomodulin (sTM), angiopoietin-2 (Ang-2), von Willebrand factor (vWF), heparan sulfate (HS), syndecan-1 (SDC-1)] levels in serum were determined by enzyme-linked immunosorbent assay (ELISA). The length of intensive care unit (ICU) stay of patients in the two groups was recorded, and the outcomes of patients in the two groups were followed up for 28 days. Results:Finally, 20 patients were enrolled in the standard treatment group, and 19 patients were enrolled in the CBP treatment group. There were no significant differences in gender, age and infection site between the two groups. The length of ICU stay in the standard treatment group was (10±5) days, and 5 patients died and 15 patients survived after 28 days. The length of ICU stay in the CBP treatment group was (9±4) days, and 8 patients died and 11 patients survived after 28 days. There were no significant differences in the length of ICU stay and number of patients who died within 28 days between the two groups (both P > 0.05). There were no significant differences in the Lac, PCT, LYM, APACHEⅡ score, SOFA score and immune function and endothelial cell injury related indexes before treatment and 1 day after treatment between the two groups. After 3 days of treatment, the Lac, PCT, APACHEⅡ score and SOFA score of the CBP treatment group were significantly lower than those before treatment, and pro-inflammatory and anti-inflammatory cytokines such as IFN-γ and IL-4, apoptosis-related indicators such as PD-1 and IL-7, and endothelial injury related factors such as sTM, SDC-1 and HS were significantly improved compared with the pre-treatment, the improvement degree of the above indicators was more obvious than that of the standard treatment group, and LYM was significantly higher than that of the standard treatment group (×10 9/L: 1.3±0.3 vs. 0.9±0.4, P < 0.05), IL-4, IFN-γ, IFN-γ/IL-4 ratio, IL-7, PD-1, sTM, SDC-1, HS, and Ang-2 were significantly lower than those of the standard treatment group [IL-4 (ng/L): 2.8 (1.5, 3.2) vs. 3.3 (2.7, 5.2), IFN-γ (ng/L): 6.3 (5.4, 106.5) vs. 217.9 (71.4, 517.1), IFN-γ/IL-4 ratio: 3.7 (1.8, 70.3) vs. 59.1 (18.3, 124.9), IL-7 (ng/L): 4.6 (3.2, 5.1) vs. 6.3 (5.2, 8.0), PD-1 (μg/L): 0.04 (0.03, 0.06) vs. 0.08 (0.05, 0.12), sTM (μg/L): 4.9 (4.3, 7.4) vs. 8.7 (6.0, 10.8), SDC-1 (μg/L): 0.6 (0.3, 1.1) vs. 0.9 (0.8, 2.5), HS (ng/L): 434.8 (256.2, 805.0) vs. 887.9 (620.1, 957.3), Ang-2 (ng/L): 934.0 (673.3, 1 502.1) vs. 2 233.9 (1 472.5, 3 808.4)], the differences were statistically significant (all P < 0.05). Conclusion:CBP treatment can eliminate the patient's immunosuppressive state, reduce a variety of endothelial injury markers and the degradation of glycocalyx, but cannot decrease the 28-day death risk or shorten the length of ICU stay.

Objective:To investigate the prognostic value of proprotein convertase subtilisin/kexin type 9 (PCSK9) and blood lipid indexes in patients with sepsis.Methods:Patients with sepsis or septic shock who were ≥ 18 years old and met the Sepsis-3.0 diagnostic criteria admitted to the department of critical care medicine of Binzhou Medical University Hospital from January to October 2021 were enrolled. Healthy adults at the same period were selected as healthy control group. Baseline characteristics, acute physiology and chronic health evaluation Ⅱ (APACHE Ⅱ) and sequential organ failure assessment (SOFA) score were recorded. Venous blood samples were collected within 24 hours after diagnosis, and serum PCSK9 was determined by enzyme-linked immunosorbent assay (ELISA) at 1, 3 days and 5 days. Meanwhile, high density lipoprotein cholesterol (HDL-C), low density lipoprotein cholesterol (LDL-C), total cholesterol (TC), triglyceride (TG) and lipoprotein A were detected. The differences of each index between sepsis group (28-day death group and survival group) and healthy control group were compared. Meanwhile, the indexes of patients with different severity and 28-day prognosis in sepsis group were compared. Receiver operator characteristic curve (ROC curve) was drawn to evaluate the predictive value of PCSK9 and blood lipid for the prognosis of sepsis. Multivariate Logistic regression was used to analyze the influencing factors for the prognosis of sepsis, and the Kaplan-Meier survival curve at 28th day was drawn.Results:There were 50 patients in sepsis group (including 19 patients with sepsis, 31 patients with septic shock) and 27 patients in healthy control group. In the sepsis group, 19 patients died and 31 patients survived within 28 days. The serum PCSK9 in the sepsis group was significantly higher than that in the healthy control group [μg/L: 223.09 (198.47, 250.82) vs. 188.00 (165.27, 214.90), P < 0.01], and HDL-C, LDL-C, TC and lipoprotein A were significantly lower than those in the healthy control group [HDL-C (mmol/L): 0.82±0.35 vs. 1.45±0.24, LDL-C (mmol/L): 1.53 (1.14, 2.47) vs. 2.89 (2.55, 3.19), TC (mmol/L): 2.03 (1.39, 2.84) vs. 4.24 (3.90, 4.71), lipoprotein A (g/L): 8.80 (5.66, 17.56) vs. 27.03 (14.79, 27.03), all P < 0.01]. PCSK9 in the sepsis death group was higher than that in the survival group [μg/L: 249.58 (214.90, 315.77) vs. 207.01 (181.50, 244.95), P < 0.01], and the HDL-C, LDL-C and TC were lower than those in the survival group [HDL-C (mmol/L): 0.64±0.35 vs. 0.93±0.30, LDL-C (mmol/L): 1.32±0.64 vs. 2.08±0.94, TC (mmol/L): 1.39 (1.01, 2.23) vs. 2.69 (1.72, 3.81), all P < 0.01]. With the progression of the disease, the PCSK9 in the sepsis death group and the survival group was significantly lower than that within 1 day of diagnosis (all P < 0.05). ROC curve analysis showed that PCSK9 had higher predictive value of 28-day death than HDL-C, LDL-C, TC [area under ROC curve (AUC) and 95% confidence interval (95% CI): 0.748 (0.611-0.885) vs. 0.710 (0.552-0.868), 0.721 (0.575-0.867), 0.702 (0.550-0.854)]. Multivariate Logistic regression analysis showed that PCSK9 was an independent risk factor affecting the 28-day prognosis of sepsis (β value was 1.014, P = 0.020). Kaplan-Meier survival curve analysis showed that when PCSK9 ≥ 208.97 μg/L, with the increase of PCSK9, the 28-day survival rate of sepsis patients decreased significantly. Conclusions:PCSK9, HDL-C, LDL-C and TC can all predict the 28-day prognosis of patients with sepsis. The prognostic value of PCSK9 is the highest. PCSK9 is an independent risk factor affecting the prognosis of sepsis. In the early stage of the disease, PCSK9 may have a good predictive value for the prognosis of sepsis. When PCSK9 ≥ 208.97 μg/L, the 28-day survival rate decreased significantly.

Objective:To explore the value of high mobility group box 1 (HMGB1), von Willebrand factor (vWF) and other cytokines in predicting the severity and prognosis of sepsis patients.Methods:Patients with sepsis and septic shock who ≥18 years old and met the Sepsis-3 diagnostic criteria admitted to the department of critical care medicine of Binzhou Medical University Hospital from January to June 2019 were taken as the research objects. The healthy individuals for regular health examination in the same period were taken as the control. The basic information, acute physiology and chronic health evaluation Ⅱ (APACHEⅡ) and sequential organ failure assessment (SOFA) scores were recorded. The venous blood was taken within 24 hours after the patients were diagnosed. The levels of HMGB1, vWF, tumour necrosis factor-α (TNF-α), interleukin-10 (IL-10), soluble thrombomodulin (sTM), vascular endothelial growth factor receptor 2 (VEGFR-2), angiopoetin-2 (Ang-2) and other cytokines in serum were determined by enzyme linked immunosorbent assay (ELISA). Differences among patients with sepsis, septic shock, healthy physical examinees, and patients who died in 28-day and those who survived, were compared. Spearman rank correlation method was used to analyze the correlation among each cytokine and APACHEⅡ, SOFA scores. The receiver operating characteristic (ROC) curve was drawn to evaluate the predictive value of cytokines on the prognosis of patients with sepsis/septic shock. Logistic regression was used to analyze the risk factors of 28-day death.Results:Eleven patients with sepsis, 25 patients with septic shock and 30 healthy individuals were enrolled. Among the patients with sepsis/septic shock, 15 died in 28-day and 21 survived. The serum levels of TNF-α, IL-10, HMGB1, vWF, sTM and VEGFR-2 in patients with sepsis were significantly higher than those in the healthy control group. The levels of TNF-α, IL-10, HMGB1, vWF, sTM in septic shock group were higher than those in the sepsis group, while the Ang-2 level decreased significantly. The serum levels of TNF-α, IL-10, HMGB1, vWF and sTM in the death group were higher than those in the survival group, while Ang-2 was lower than the survival group. Spearman correlation analysis showed that HMGB1, TNF-α, sTM, IL-10, vWF were positively correlated with APACHEⅡ score when patients with sepsis/septic shock were enrolled ( r values were 0.652, 0.666, 0.445, 0.430 and 0.355, respectively, all P < 0.05), and HMGB1, TNF-α also positively correlated with SOFA score ( r values were 0.433, 0.479, both P < 0.05). Ang-2 was negatively correlated with APACHEⅡ and SOFA scores ( r values were -0.519, -0.440, both P < 0.05). ROC curve analysis showed that the predictive value of HMGB1, vWF, IL-10, sTM for 28-day death in patients with sepsis/septic shock were higher than the APACHEⅡ score [the area under ROC curve (AUC) and 95% confidence interval (95% CI): 0.946 (0.870-1.000), 0.902 (0.790-1.000), 0.877 (0.745-1.000), 0.868 (0.734-1.000) vs. 0.846 (0.700-0.991)]. Logistic regression analysis showed that APACHEⅡ score, vWF, sTM, and IL-10 were independent risk factors for 28-day death in patients with sepsis/septic shock (β values were 4.731, 0.407, -7.058, -0.887, all P < 0.05). Conclusion:HMGB1, vWF, IL-10, sTM and other cytokines all can be used to evaluate the severity and prognosis of sepsis patients.