A new deep learning model for assisted diagnosis on electrocardiogram.

In order to enhance the accuracy of computer aided electrocardiogram analysis, we propose a deep learning model called CBRNN to assist diagnosis on electrocardiogram for clinical medical service. It combines two sub networks which are convolutional neural network (CNN) and bi-directional recurrent neural network (BRNN). In the model, CNN with one-dimension convolution is employed to extract features for each lead of ECG, and BRNN is used to fuse features of different leads to represent deeper features. In the training step, we use more than 40 thousand training data and more than 19 thousand validation data to obtain the optimal parameters of the model. Besides, by validating our model on more than CCDD 120,000 real data, it achieves an 87.69% accuracy rate, higher than popular deep learning models such as CNN and ResNet. Our model has better accuracy than state-of-the-art models and it is also slightly higher than the average accuracy of human judgement. It can be served for the first round screening of ECG examination clinical diagnosis.

Fluorescent microsatellite analysis of urine sediment in patients with urothelial carcinoma.

AIMS: To detect microsatellite loci alterations by fluorescent multiplex PCR in urine sediment cell of urothelial carcinoma, and to determine if they can be used as genetic markers for diagnosis of urothelial carcinoma. MATERIALS AND METHODS: Microsatellite alteration analysis was conducted using fluorescent multiplex PCR with samples from 64 cases of urothelial carcinomas of the bladder. Three microsatellites spanning the 3p14 and two additional microsatellites in 9q33 and 9p22 were analyzed. Microsatellite alterations (microsatellite instability and/or loss of heterozygosity) in urine sediment cells of urothelial carcinoma patients matched for peripheral blood and tumor tissue were all analyzed. RESULTS: The frequency of microsatellite alterations in urothelial carcinoma was chromosome 3p (D3S1234: 14.6% (7/48), D3S1300: 16.7% (8/48), D3S1313: 8.35% (4/48)); 9q (D9S242: 33.3% (16/48)), and 9p (D9S162: 27.1% (13/48)). Microsatellite alterations happened in 62.5% (40/64) of the patients when combined with all five markers. Our study showed a significant correlation between the microsatellite alteration of the five-locus panel and recurrence (p = 0.010) and smoking habit (p = 0.006). CONCLUSIONS: The results suggest that these microsatellite loci alterations may have an important role in the recurrence of urothelial carcinomas. Further studies are needed to better determine the effect of microsatellite loci alterations on prognosis.

Relationship and prognostic significance of SPARC and VEGF protein expression in colon cancer.

BACKGROUND: SPARC (secreted protein, acidic and rich in cysteine) is closely related with the progress, invasion and metastasis of malignant tumor and angiogenesis. METHODS: Using human colon adenocarcinoma tissues (hereinafter referred to as colon cancer) and their corresponding non-diseased colon from 114 patients' biopsies, the expression of SPARC and vascular endothelial growth factor (VEGF) were investigated by immunohistochemistry staining to assessment the relationship between SPARC and VEGF, as well as their prognostic significance in patients. Evaluation of VEGF expression level with the same tissues was used to establish the antigenic profiles, and the marker of CD34 staining was used as an indicator of microvessel density (MVD). RESULTS: SPARC expression was mainly in the stromal cells surrounding the colon cancer, and was significant difference in those tissues with the lymph node metastasis and differentiation degree of tumor. Expression of SPARC was significantly correlated with the expression of VEGF and MVD in colon cancer tissues. Patients with low or absence expressing SPARC had significantly worse overall survival and disease-free survival in a Single Factor Analysis; Cox Regression Analysis, SPARC emerged as an overall survival and disease-free survival independent prognostic factor for colon cancer. CONCLUSION: The low expression or absence of stromal SPARC was an independent prognostic factor for poor prognosis of colon cancer. SPARC maybe involved in the regulation of anti-angiogenesis by which it may serve as a novel target for colon cancer treatment as well as a novel distinctive marker.

Dysadherin expression in gastrointestinal stromal tumors (GISTs).

The purpose of this study was to detect the protein and mRNA expression of dysadherin and to investigate the clinical significance of dysadherin expression in gastrointestinal stromal tumors (GISTs). Twenty fresh GISTs samples were used for testing the mRNA of dysadherin and E-cadherin by reverse-transcription polymerase chain reaction (RT-PCR). In addition, 54 paraffin blocks of GISTs cases were also used for the investigation of dysadherin and E-cadherin using immunohistochemistry. The dysadherin expression level was significantly correlated with GISTs risk stratification (chi(2)=5.769, P<0.05), but was not related to age, sex, histologic subtype, and location (P>0.05). There was only faint or absent E-cadherin protein expression in GISTs. The expression level of dysadherin is related to the risk of GISTs. Dysadherin upregulation, which may cause loss of E-cadherin, is one of the reasons for GISTs recurrence and metastasis. It seems that dysadherin protein detection is a promising method for GISTs prognostication.

Quantitative structure activity relationship models based on heuristic method and gene expression programming for the prediction of the pK(a) values of sulfa drugs / 药学学报

Quantitative structure-property relationships (QSPR) were developed to predict the pK(a) values of sulfa drugs via heuristic method (HM) and gene expression programming (GEP). The descriptors of 31 sulfa drugs were calculated by the software CODESSA, which can calculate constitutional, topological, geometrical, electrostatic, and quantum chemical descriptors. HM was also used for the preselection of 4 appropriate molecular descriptors. Linear and nonlinear QSPR models were developed based on the HM and GEP separately and two prediction models lead to a good correlation coefficient (R) of 0.90 and 0.95. The two QSPR models are tseful in predicting pK(a) during the discovery of new drugs and providing theory information for studying the new drugs.

[Prognostic value of P16 gene methylation and P16 protein expression in gastrointestinal stromal tumor].

OBJECTIVE: To study the methylation status of P16 gene promoter and the expression of P16 protein in gastrointestinal stromal tumors(GIST) and to explore the prognostic value. METHODS: Methylation status of the P16 promoter was detected by methylation- specific polymerase chain reaction (MSP) and the expression of P16 protein by immunohistochemistry in 62 patients with GIST. RESULTS: The status of P16 gene methylation and the expression of P16 protein were significantly different among the patients with different subclassification of GIST using Fletcher's scheme (P < 0.05, P < 0.01). And there were significant differences in progressive disease (PD) among various levels of P16 expression (P < 0.01). P16 gene methylation was closely related to P16 protein. P16 gene methylation accounted for 75% in the tumor tissue with less than 50% P16 positive cells, and accounted for only 10% in the tumor tissue with more than 50% P16 positive cells (P< 0.01). CONCLUSION: P16 immunohistochemical assessment can be used as a prognostic index for GIST. The patients with more than 50% fraction of cells with low P16 immunostaining have poor prognosis.