OBJECTIVE: To develop a few-shot learning (FSL) approach for classifying optical coherence tomography (OCT) images in patients with inherited retinal disorders (IRDs). METHODS: In this study, an FSL model based on a student-teacher learning framework was designed to classify images. 2,317 images from 189 participants were included. Of these, 1,126 images revealed IRDs, 533 were normal samples, and 658 were control samples. RESULTS: The FSL model achieved a total accuracy of 0.974-0.983, total sensitivity of 0.934-0.957, total specificity of 0.984-0.990, and total F1 score of 0.935-0.957, which were superior to the total accuracy of the baseline model of 0.943-0.954, total sensitivity of 0.866-0.886, total specificity of 0.962-0.971, and total F1 score of 0.859-0.885. The performance of most subclassifications also exhibited advantages. Moreover, the FSL model had a higher area under curves (AUC) of the receiver operating characteristic (ROC) curves in most subclassifications. CONCLUSION This study demonstrates the effective use of the FSL model for the classification of OCT images from patients with IRDs, normal, and control participants with a smaller volume of data. The general principle and similar network architectures can also be applied to other retinal diseases with a low prevalence.
Humans ; Tomography, Optical Coherence ; Deep Learning ; Retinal Diseases/diagnostic imaging* ; Retina/diagnostic imaging* ; ROC Curve

OBJECTIVES: This study aims to predict the risk of deep caries exposure in radiographic images based on the convolutional neural network model, compare the prediction results of the network model with those of senior dentists, evaluate the performance of the model for teaching and training stomatological students and young dentists, and assist dentists to clarify treatment plans and conduct good doctor-patient communication before surgery. METHODS: A total of 206 cases of pulpitis caused by deep caries were selected from the Department of Stomatological Hospital of Tianjin Medical University from 2019 to 2022. According to the inclusion and exclusion criteria, 104 cases of pulpitis were exposed during the decaying preparation period and 102 cases of pulpitis were not exposed. The 206 radiographic images collected were randomly divided into three groups according to the proportion: 126 radiographic images in the training set, 40 radiographic images in the validation set, and 40 radiographic images in the test set. Three convolutional neural networks, visual geometry group network (VGG), residual network (ResNet), and dense convolutional network (DenseNet) were selected to analyze the rules of the radiographic images in the training set. The radiographic images of the validation set were used to adjust the super parameters of the network. Finally, 40 radiographic images of the test set were used to evaluate the performance of the three network models. A senior dentist specializing in dental pulp was selected to predict whether the deep caries of 40 radiographic images in the test set were exposed. The gold standard is whether the pulp is exposed after decaying the prepared hole during the clinical operation. The prediction effect of the three network models (VGG, ResNet, and DenseNet) and the senior dentist on the pulp exposure of 40 radiographic images in the test set were compared using receiver operating characteristic (ROC) curve, area under the ROC curve (AUC), accuracy, sensitivity, specificity, positive predictive value, negative predictive value, and F1 score to select the best network model. RESULTS: The best network model was DenseNet model, with AUC of 0.97. The AUC values of the ResNet model, VGG model, and the senior dentist were 0.89, 0.78, and 0.87, respectively. Accuracy was not statistically different between the senior dentist (0.850) and the DenseNet model (0.850)(P>0.05). Kappa consistency test showed moderate reliability (Kappa=0.6>0.4, P<0.05). CONCLUSIONS Among the three convolutional neural network models, the DenseNet model has the best predictive effect on whether deep caries are exposed in imaging. The predictive effect of this model is equivalent to the level of senior dentists specializing in dental pulp.
Humans ; Deep Learning ; Neural Networks, Computer ; Pulpitis/diagnostic imaging* ; Reproducibility of Results ; ROC Curve ; Random Allocation

Background A simple measurement of central venous pressure (CVP)-mean by the digital monitor display has become increasingly popular. However, the agreement between CVP-mean and CVP-end (a standard method of CVP measurement by analyzing the waveform at end-expiration) is not well determined. This study was designed to identify the relationship between CVP-mean and CVP-end in critically ill patients and to introduce a new parameter of CVP amplitude (ΔCVP= CVPmax - CVPmin) during the respiratory period to identify the agreement/disagreement between CVP-mean and CVP-end.Methods In total, 291 patients were included in the study. CVP-mean and CVP-end were obtained simultaneously from each patient. CVP measurement difference (|CVP-mean - CVP-end|) was defined as the difference between CVP-mean and CVP-end. The ΔCVP was calculated as the difference between the peak (CVPmax) and the nadir value (CVPmin) during the respiratory cycle, which was automatically recorded on the monitor screen. Subjects with |CVP-mean - CVP-end|≥ 2 mmHg were divided into the inconsistent group, while subjects with |CVP-mean - CVP-end| < 2 mmHg were divided into the consistent group.Results ΔCVP was significantly higher in the inconsistent group [7.17(2.77) vs.5.24(2.18), P<0.001] than that in the consistent group. There was a significantly positive relationship between ΔCVP and |CVP-mean - CVP-end| (r=0.283, P <0.0001). Bland-Altman plot showed the bias was -0.61 mmHg with a wide 95% limit of agreement (-3.34, 2.10) of CVP-end and CVP-mean. The area under the receiver operating characteristic curves (AUC) of ΔCVP for predicting |CVP-mean - CVP-end| ≥ 2 mmHg was 0.709. With a high diagnostic specificity, using ΔCVP<3 to detect |CVP-mean - CVP-end| lower than 2mmHg (consistent measurement) resulted in a sensitivity of 22.37% and a specificity of 93.06%. Using ΔCVP>8 to detect |CVP-mean - CVP-end| >8 mmHg (inconsistent measurement) resulted in a sensitivity of 31.94% and a specificity of 91.32%.Conclusions CVP-end and CVP-mean have statistical discrepancies in specific clinical scenarios. ΔCVP during the respiratory period is related to the variation of the two CVP methods. A high ΔCVP indicates a poor agreement between these two methods, whereas a low ΔCVP indicates a good agreement between these two methods.
Humans ; Central Venous Pressure ; Respiration ; ROC Curve

OBJECTIVES: To study the changes in cell free-DNA (cf-DNA), a marker of neutrophil extracellular traps (NETs), in neonates with acute respiratory distress syndrome (ARDS), and to evaluate its relationship with the severity and early diagnosis of ARDS. METHODS: The neonates diagnosed with ARDS in the Affiliated Hospital of Jiangsu University from January 2021 to June 2022 were enrolled in the prospective study. The neonates were divided into mild, moderate, and severe ARDS groups based on the oxygen index (OI) (4≤OI<8, 8≤OI<16, and OI≥16, respectively). The control group was selected from jaundice neonates who were observed in the neonatal department of the hospital during the same period, and they had no pathological factors causing neonatal jaundice. Peripheral blood samples were collected on day 1, day 3, and day 7 after admission for the ARDS group, and on the day of admission for the control group. Serum cf-DNA levels were measured using a fluorescence enzyme-linked immunosorbent assay. Serum interleukin-6 (IL-6) and tumor necrosis factor-α (TNF-α) levels were measured using enzyme-linked immunosorbent assay. A Pearson correlation analysis was used to evaluate the correlation of serum cf-DNA levels with serum IL-6 and TNF-α levels. RESULTS: A total of 50 neonates were enrolled in the ARDS group, including 15 neonates with mild ARDS, 25 with moderate ARDS, and 10 with severe ARDS. Twenty-five neonates were enrolled in the control group. Compared with the control group, the serum levels of cf-DNA, IL-6, and TNF-α in all ARDS groups were significantly increased (P<0.05). Compared with the mild ARDS group, the serum levels of cf-DNA, IL-6, and TNF-α in the moderate and severe ARDS groups were significantly increased (P<0.05), and the increase was more significant in the severe ARDS group (P<0.05). The serum levels of cf-DNA, IL-6, and TNF-α in all ARDS groups were significantly increased on day 3 after admission and significantly decreased on day 7 after admission compared with those on day 1 after admission (P<0.05). The Pearson correlation analysis showed that there was a positive correlation between serum cf-DNA levels and IL-6 levels as well as TNF-α levels in 50 neonates with ARDS (P<0.05). CONCLUSIONS There is an excessive expression of NETs in neonates with ARDS, and dynamic monitoring of serum cf-DNA levels has certain clinical value in evaluating the severity and early diagnosis of ARDS in neonates.
Infant, Newborn ; Humans ; Extracellular Traps ; Prospective Studies ; Tumor Necrosis Factor-alpha ; Interleukin-6 ; Prognosis ; ROC Curve ; Respiratory Distress Syndrome ; DNA

UNLABELLED: AbstractObjective: To analysis the clinical data of patients after single-center hematopoietic stem cell transplantation (HSCT) and construct a predictive model for metabolic syndrome (MS) diagnosis. METHODS: Ninety-three hematology patients who underwent HSCT at the First Hospital of Lanzhou University from July 2015 to September 2022 were selected to collect basic data, transplantation status and postoperative data, the clinical characteristics of patients with and without MS after transplantation were compared and analyzed. Logistic regression model was used to analyze the influence fators of MS after transplantation, and a predictive model of HSCT-MS diagnosis was constructed under the influence of independent influence factors. The model was evaluated using the ceceiver operating characteristic curve (ROC curve). RESULTS: Metabolic syndrome occurred in 36 of 93 HSCT patients and did not occur in 57. Compared with non-HSCT-MS group, HSCT-MS had significantly higher fasting blood glucose (FBG) levels before transplantation, shorter course before transplantation, and higher bilirubin levels after transplantation (P<0.05). The statistically significant clinical indicators were subjected to multi-factor logistic regression analysis, and the results showed that pre-transplant high FBG, pre-transplant short disease course and post-transplant high bilirubin were independent influence factors for HSCT-MS. The standard error of predicting the occurrence of HSCT-MS based on the clinical model was 0.048, the area under the curve AUC=0.776, 95% CI :0.683-0.869, the optimal threshold was 0.58 based on the Jorden index at maximum, the sensitivity was 0.694, and the specificity was 0.772, which has certain accuracy. CONCLUSION A clinical prediction model for HSCT-MS based on logistic regression analysis is constructed through the analysis of clinical data, which has certain clinical value.
Humans ; Metabolic Syndrome ; Prognosis ; Models, Statistical ; Hematopoietic Stem Cell Transplantation ; ROC Curve ; Retrospective Studies

OBJECTIVE: To investigate the expression and clinical significance of soluble Fas (sFas) and sFasL in patients with secondary hemophagocytic lymphohistiocytosis (sHLH). METHODS: From September 2015 to December 2020, 86 sHLH patients who met the HLH2004 diagnostic criteria were collected. They were divided into 55 cases in the MAHLH group and 31 cases in the NonMAHLH group according to the etiology. Thirty healthy persons were chosen as the normal control group, and 20 patients with systemic lupus erythematosus (SLE) were chosen as the disease control group. The expression levels of sFas and sFasL in the serum of patients with each group were detected by ELISA, and the clinical data were collected for statistical analysis. The significance of sFas and sFasL in sHLH was analyzed by ROC curve. RESULTS: Serum levels of sFas and sFasL in patients with newly diagnosed sHLH were significantly higher than those in disease control group and normal control group (P<0.01). The levels of sFas and sFasL in MAHLH group were significantly higher than those in nonMAHLH (infection related HLH and autoimmune disease related HLH) group (P<0.01). The serum levels of sFas and sFasL in 17 newly treated patients with sHLH (17/86) after treatment were significantly lower than those before treatment (P<0.01). The serum sFas level in newly diagnosed sHLH patients was positively correlated with SF(r=0.35), sCD25(r=0.79) and sFasL(r=0.73). The serum sFasL level was positively correlated with SF(r=0.39), sCD25(r=0.64) and sFas(r=0.73). Compared with the NonMAHLH group, the area under the ROC curve was 0.707 (95% CI: 0.593-0.821) (P=0.0015). The optimal critical value for diagnosing MAHLH by sFas level was 12 743 pg/ml, and the sensitivity and specificity were 70.9% and 71% respectively. Compared with the NonMAHLH group, the area under the ROC curve was 0.765(95% CI: 0.659-0.87)(P<0.01). The median OS time of sFas high expression group (≥16798.5 pg/ml) and sFasL high expression group (≥4 785 pg/ml) was significantly shorter than that of the low expression group (P<0.001). CONCLUSION Serum levels of sFas and sFasL can be used for the early diagnosis and differential diagnosis of sHLH disease, and are the factor related to the poor prognosis of sHLH.
Humans ; Lymphohistiocytosis, Hemophagocytic ; Clinical Relevance ; ROC Curve ; Sensitivity and Specificity ; Lupus Erythematosus, Systemic

Objective: To compare the predictive efficacy of the two thrombosis risk assessment scores (Padua and IMPEDE scores) in venous thromboembolism (VTE) within 6 months in patients with newly diagnosed multiple myeloma (NDMM) in China. Methods: This study reviewed the clinical data of 421 patients with NDMM hospitalized in Beijing Jishuitan Hospital from April 2014 to February 2022. The sensitivity, specificity, accuracy, and Youden index of the two scores were calculated to quantify the thrombus risk assessment of VTE by the Padua and IMPEDE scores. The receiver operating characteristics curves of the two evaluation scores were drawn. Results: The incidence of VTE was 14.73%. The sensitivity, specificity, accuracy, and Youden index of the Padua score were 100%, 0%, 14.7%, and 0% and that of the IMPEDE score was 79%, 44%, 49.2%, and 23%, respectively. The areas under the curve of Padua and IMPEDE risk assessment scores were 0.591 and 0.722, respectively. Conclusion: IMPEDE score is suitable for predicting VTE within 6 months in patients with NDMM.
Humans ; Venous Thromboembolism/etiology* ; Multiple Myeloma/diagnosis* ; Risk Assessment ; Risk Factors ; ROC Curve ; Retrospective Studies

Objective: To investigate the feasibility and value of histogram analysis based on two-dimensional gray-scale ultrasonography in the differential diagnosis of medullary thyroid carcinoma (MTC) and thyroid adenoma (TA). Methods: The preoperative ultrasound images of 86 newly diagnosed MTC patients and 100 TA patients treated in the Cancer Hospital of Chinese Academy of Medical Sciences from January 2015 to October 2021 were collected. Histograms were performed based on the regions of interest (ROIs) delineated manually by two radiologists, thereafter, mean, variance, skewness, kurtosis, percentiles (1st, 10th, 50th, 90th, 99th) were generated. The histogram parameters between the MTC group and the TA group were compared, and the independent predictors were screened by multivariate logistic regression analysis. Receiver operating characteristic (ROC) analysis was used to compare the individual diagnostic efficacy and joint diagnostic efficacy of independent predictors. Results: Multivariate regression analysis showed that mean, skewness, kurtosis and 50th percentile were independent factors. The skewness and kurtosis in the MTC group were significantly higher than those in the TA group, and the mean and 50th percentile were significantly lower than those in the TA group. The area under the individual ROC curve of mean, skewness, kurtosis and 50th percentile is 0.654-0.778. The area under the combined ROC curve is 0.826. Conclusion: Histogram analysis based on two-dimensional gray-scale ultrasonography is a promising tool to distinguish MTC from TA, in which the joint diagnosis value of mean, skewness, kurtosis and 50th percentile is the highest.
Humans ; ROC Curve ; Diagnosis, Differential ; Retrospective Studies ; Thyroid Neoplasms/diagnostic imaging* ; Ultrasonography ; Diffusion Magnetic Resonance Imaging/methods*

Objective: To determine the ability of the ratio of platelet to lymphocyte (PLR) for predicting extubation failure in septic patients receiving invasive mechanical ventilation (IMV). Methods: The retrospective cohort study was conducted in ICU at Beijing Chao-Yang Hospital Shijingshan District, Capital Medical University in China from January, 2010 to December, 2019, including patients with sepsis who received IMV. 180 patients were enrolled in the study, including 111 male and 69 female, with the age range of 23-93 years and the median age of 76 years, and with an average age of 71.22 years. The medical records were reviewed, such as age, sex, acute physiology and chronic health evaluation II (APACHEII), sequential organ failure assessment (SOFA), spontaneous breathing trial (SBT) outcome, weaning outcome, complete blood count before SBT. According to weaning outcome, patients were divided into weaning failure and weaning success group. The difference of PLR, white blood cell(WBC), C-reaction protein (CRP) and procalcitonin (PCT) were compared between weaning failure and success group. Receiver-operating characteristics (ROC) curves and multivariate logistical regression analysis were employed to analyze the performance of these inflammatory markers for predicting weaning failure in patients with sepsis. Results: 180 patients with sepsis were included in the study and 37 patients (20.5%) experienced weaning failure (31 SBT failure and 6 extubation failure after successful SBT). PLR was higher in weaning failure group than that in weaning success group (Z=-5.793,P<0.001). Other inflammation biomarkers such as WBC, CRP and PCT were also higher in weaning failure group than that in weaning success group(Z=-4.356, -3.118 and -2.743, P<0.001, 0.002 and 0.006, respectively). According to ROC curves, PLR has a better predictive ability for weaning failure (AUC=0.809,95%CI 0.733-0.885) when compared to WBC (AUC=0.773,95%CI 0.648-0.817), CRP (AUC=0.666,95%CI 0.577-0.755) and PCT (AUC=0.603,95%CI 0.508-0.698). The cutoff value of PLR for predicting weaning failure was 257.69 with sensitivity 78.38%, specificity 76.22%, and diagnostic accuracy 71.66%. According to multivariate logistic regression analyses, PLR>257.69 was an independent risk factor for predicting weaning failure in patients with sepsis. Conclusions: PLR may be a valuable biomarker for predicting weaning failure in septic patients receiving IMV, and the patients with higher PLR should be handled with caution since they are at higher risk of weaning failure, and some more effective treatment should be in consideration after extubation.
Humans ; Male ; Female ; Aged ; Young Adult ; Adult ; Middle Aged ; Aged, 80 and over ; Respiration, Artificial ; Retrospective Studies ; Sepsis/diagnosis* ; Procalcitonin ; C-Reactive Protein ; Biomarkers ; ROC Curve ; Lymphocytes

Objective: Compare and analyze the results of the domestic Lanyi AH600 glycated hemoglobin analyzer and other different detection systems to understand the comparability of the detection results of different detectors, and establish the best cut point of Lanyi AH600 determination of haemoglobin A1c (HbA1c) in the diagnosis of diabetes. Methods: Multi center cohort study was adopted. The clinical laboratory departments of 18 medical institutions independently collected test samples from their respective hospitals from March to April 2022, and independently completed comparative analysis of the evaluated instrument (Lanyi AH600) and the reference instrument HbA1c. The reference instruments include four different brands of glycosylated hemoglobin meters, including Arkray, Bio-Rad, DOSOH, and Huizhong. Scatter plot was used to calculate the correlation between the results of different detection systems, and the regression equation was calculated. The consistency analysis between the results of different detection systems was evaluated by Bland Altman method. Consistency judgment principles: (1) When the 95% limits of agreement (95% LoA) of the measurement difference was within 0.4% HbA1c and the measurement score was≥80 points, the comparison consistency was good; (2) When the measurement difference of 95% LoA exceeded 0.4% HbA1c, and the measurement score was≥80 points, the comparison consistency was relatively good; (3) The measurement score was less than 80 points, the comparison consistency was poor. The difference between the results of different detection systems was tested by paired sample T test or Wilcoxon paired sign rank sum test; The best cut-off point of diabetes was analyzed by receiver operating characteristic curve (ROC). Results: The correlation coefficient R² of results between Lanyi AH600 and the reference instrument in 16 hospitals is≥0.99; The Bland Altman consistency analysis showed that the difference of 95% LoA in Nanjing Maternity and Child Health Care Hospital in Jiangsu Province (reference instrument: Arkray HA8180) was -0.486%-0.325%, and the measurement score was 94.6 points (473/500); The difference of 95% LoA in the Tibetan Traditional Medical Hospital of TAR (reference instrument: Bio-Rad Variant II) was -0.727%-0.612%, and the measurement score was 89.8 points; The difference of 95% LoA in the People's Hospital of Chongqing Liang Jiang New Area (reference instrument: Huizhong MQ-2000PT) was -0.231%-0.461%, and the measurement score was 96.6 points; The difference of 95% LoA in the Taihe Hospital of traditional Chinese Medicine in Anhui Province (reference instrument: Huizhong MQ-2000PT) was -0.469%-0.479%, and the measurement score was 91.9 points. The other 14 hospitals, Lanyi AH600, were compared with 4 reference instrument brands, the difference of 95% LoA was less than 0.4% HbA1c, and the scores were all greater than 95 points. The results of paired sample T test or Wilcoxon paired sign rank sum test showed that there was no statistically significant difference between Lanyi AH600 and the reference instrument Arkray HA8180 (Z=1.665,P=0.096), with no statistical difference. The mean difference between the measured values of the two instruments was 0.004%. The comparison data of Lanyi AH600 and the reference instrument of all other institutions had significant differences (all P<0.001), however, it was necessary to consider whether it was within the clinical acceptable range in combination with the results of the Bland-Altman consistency analysis. The ROC curve of HbA1c detected by Lanyi AH600 in 985 patients with diabetes and 3 423 patients with non-diabetes was analyzed, the area under curve (AUC) was 0.877, the standard error was 0.007, and the 95% confidence interval 95%CI was (0.864, 0.891), which was statistically significant (P<0.001). The maximum value of Youden index was 0.634, and the corresponding HbA1c cut point was 6.235%. The sensitivity and specificity of diabetes diagnosis were 76.2% and 87.2%, respectively. Conclusion: Among the hospitals and instruments currently included in this study, among these four hospitals included Nanjing Maternity and Child Health Care Hospital in Jiangsu Province (reference instrument: Arkray HA8180), Tibetan Traditional Medical Hospital of TAR (reference instrument: Bio-Rad Variant Ⅱ), the People's Hospital of Chongqing Liang Jiang New Area (reference instrument: Huizhong MQ-2000PT), and the Taihe Hospital of traditional Chinese Medicine in Anhui Province (reference instrument: Huizhong MQ-2000PT), the comparison between Lanyi AH600 and the reference instruments showed relatively good consistency, while the other 14 hospitals involved four different brands of reference instruments: Arkray, Bio-Rad, DOSOH, and Huizhong, Lanyi AH600 had good consistency with its comparison. The best cut point of the domestic Lanyi AH600 for detecting HbA1c in the diagnosis of diabetes is 6.235%.
Pregnancy ; Child ; Humans ; Female ; Glycated Hemoglobin ; Cohort Studies ; Diabetes Mellitus/diagnosis* ; Sensitivity and Specificity ; ROC Curve