Comparing deep neural network and other machine learning algorithms for stroke prediction in a large-scale population-based electronic medical claims database.

Electronic medical claims (EMCs) can be used to accurately predict the occurrence of a variety of diseases, which can contribute to precise medical interventions. While there is a growing interest in the application of machine learning (ML) techniques to address clinical problems, the use of deep-learning in healthcare have just gained attention recently. Deep learning, such as deep neural network (DNN), has achieved impressive results in the areas of speech recognition, computer vision, and natural language processing in recent years. However, deep learning is often difficult to comprehend due to the complexities in its framework. Furthermore, this method has not yet been demonstrated to achieve a better performance comparing to other conventional ML algorithms in disease prediction tasks using EMCs. In this study, we utilize a large population-based EMC database of around 800,000 patients to compare DNN with three other ML approaches for predicting 5-year stroke occurrence. The result shows that DNN and gradient boosting decision tree (GBDT) can result in similarly high prediction accuracies that are better compared to logistic regression (LR) and support vector machine (SVM) approaches. Meanwhile, DNN achieves optimal results by using lesser amounts of patient data when comparing to GBDT method.

Association between Fibrinogen and Leukoaraiosis in Patients with Ischemic Stroke and Atrial Fibrillation.

BACKGROUND: Leukoaraiosis (LA), a surrogate of cerebral small-vessel diseases (CSVD), has been increasingly recognized because of its high prevalence and strong prognostic value in stroke. But the mechanism of LA is incompletely clarified. Fibrinogen is a crucial role in coagulation cascade and inflammation. There are inconsistent reports on the association of fibrinogen with LA in the general population. We aimed to investigate the association between fibrinogen and LA in patients with stroke and atrial fibrillation (AF), which was not ever reported before. METHODS: Patients with ischemic stroke and AF were prospectively and consecutively recruited. Clinico-demographic data and fibrinogen levels were collected within 48 hours from stroke onsets and analyzed according to the presence and distribution of LA (periventricular hyperintensity [PVH] and deep white matter hyperintensity). RESULTS: Of 186 patients (34.4% male; mean age, 68.76 ± 12.76 years) enrolled, 134 patients (72.0%) presented with LA. Elevated fibrinogen levels were associated with higher presence of LA (P = .005) and PVH (P = .002). After adjustment for the confounders, the fibrinogen levels were independently correlated with LA and PVH (all P <.05). Patients with elevated fibrinogen levels (≥3.5 g/L) were more likely to present with LA and PVH, with the odds ratios of 14.037 (95% confidence interval [CI] 2.588-76.131) and 12.567 (95% CI 2.572-61.395), respectively. CONCLUSION: This study found that fibrinogen was independently and positively associated with LA and PVH in patients with stroke and AF. These results provide further evidence for the key role of fibrinogen in LA, even the total CSVD burden.

Association between leukoaraiosis and hemorrhagic transformation after cardioembolic stroke due to atrial fibrillation and/or rheumatic heart disease.

Cardioembolic stroke due to atrial fibrillation (AF) and/or rheumatic heart disease (RHD) often involves hemorrhagic transformation (HT), and we examined whether leukoaraiosis (LA) was associated with HT in these cases. We prospectively enrolled 251 patients who were admitted to two hospitals within one month of experiencing cardioembolic stroke due to AF/RHD. LA severity was assessed using three visual rating scales. HT was identified in 99 patients (39.4%) based on baseline computed tomography (CT) and post-admission magnetic resonance imaging or second CT. Univariate analysis identified risk of HT as higher in the presence of frontal LA based on the age-related white matter changes scale and in the presence of anterior LA based on the VSS scale. Multivariate analysis confirmed that moderate to severe LA was independently associated with higher HT risk. Of the various sites affected in LA, frontal LA correlated with highest risk of HT (OR 3.199, 95%CI 1.555-6.580). These results suggest that moderate to severe LA, especially at periventricular and anterior sites, is associated with HT after cardioembolic stroke due to AF/RHD. These findings suggest the need to take LA into account as a HT risk factor when considering the use of anticoagulation and thrombolysis in these patients.

Long intergenic non-coding RNA 00152 promotes lung adenocarcinoma proliferation via interacting with EZH2 and repressing IL24 expression.

BACKGROUND: Numerous studies have shown that long non-coding RNAs (lncRNAs) behave as a novel class of transcript during multiple cancer processes, such as cell proliferation, apoptosis, migration, and invasion. LINC00152 is located on chromosome 2p11.2, and has a transcript length of 828 nucleotides. The biological role of LINC00152 in LAD(lung adenocarcinoma) remains unknown. METHODS: Quantitative reverse transcription PCR(qRT-PCR) was used to detect LINC00152 expression in 60 human LAD tissues and paired normal tissues. In vitro and in vivo studies showed the biological function of LINC00152 in tumour progression. RNA transcriptome sequencing technology was performed to identify the downstream suppressor IL24(interleukin 24) which was further examined by qRT-PCR, western bolt and rescue experiments. RNA immunoprecipitation (RIP), RNA pulldown, and Chromatin immunoprecipitation (ChIP) assays were carried out to reveal the interaction between LINC00152, EZH2 and IL24. RESULTS: LINC00152 expression was upregulated in 60 human LAD tissues and paired normal tissues. High levels of LINC00152 expression were correlated with advanced TNM stage, larger tumor size, and lymph node metastasis, as well as shorter survival time. Silencing of LINC00152 suppressed cell growth and induced cell apoptosis. LINC00152 knockdown altered the expression of many downstream genes, including IL24. LINC00152 could interact with EZH2 and inhibit IL24 transcription. Moreover, the ectopic expression of IL24 repressed cell proliferation and partly reversed LINC00152 overexpression-induced promotion of cell growth in LAD. CONCLUSIONS: Our study reveals an oncogenic role for LINC00152 in LAD tumorigenesis, suggesting that it could be used as a therapeutic target in LAD treatment.

Long non-coding RNAs in anti-cancer drug resistance.

Chemotherapy is one of the basic treatments for cancers; however, drug resistance is mainly responsible for the failure of clinical treatment. The mechanism of drug resistance is complicated because of interaction among various factors including drug efflux, DNA damage repair, apoptosis and targets mutation. Long non-coding RNAs (lncRNAs) have been a focus of research in the field of bioscience, and the latest studies have revealed that lncRNAs play essential roles in drug resistance in breast cancer, gastric cancer and lung cancer, et al. Dysregulation of multiple targets and pathways by lncRNAs results in the occurrence of chemoresistance. In this review, we will discuss the mechanisms underlying lncRNA-mediated resistance to chemotherapy and the therapeutic potential of lncRNAs in future cancer treatment.

The pseudogene DUXAP10 promotes an aggressive phenotype through binding with LSD1 and repressing LATS2 and RRAD in non small cell lung cancer.

Pseudogenes have been considered as non-functional transcriptional relics of human genomic for long time. However, recent studies revealed that they play a plethora of roles in diverse physiological and pathological processes, especially in cancer, and many pseudogenes are transcribed into long noncoding RNAs and emerging as a novel class of lncRNAs. However, the biological roles and underlying mechanism of pseudogenes in the pathogenesis of non small cell lung cancer are still incompletely elucidated. This study identifies a putative oncogenic pseudogene DUXAP10 in NSCLC, which is located in 14q11.2 and 2398 nt in length. Firstly, we found that DUXAP10 was significantly up-regulated in 93 human NSCLC tissues and cell lines, and increased DUXAP10 was associated with patients poorer prognosis and short survival time. Furthermore, the loss and gain of functional studies including growth curves, migration, invasion assays and in vivo studies verify the oncogenic roles of DUXAP10 in NSCLC. Finally, the mechanistic experiments indicate that DUXAP10 could interact with Histone demethylase Lysine specific demethylase1 (LSD1) and repress tumor suppressors Large tumor suppressor 2 (LATS2) and Ras-related associated with diabetes (RRAD) transcription in NSCLC cells. Taken together, these findings demonstrate DUXAP10 exerts the oncogenic roles through binding with LSD1 and epigenetic silencing LATS2 and RRAD expression. Our investigation reveals the novel roles of pseudogene in NSCLC, which may serve as new target for NSCLC diagnosis and therapy.

Long Noncoding RNA PVT1 Promotes Non-Small Cell Lung Cancer Cell Proliferation through Epigenetically Regulating LATS2 Expression.

Long noncoding RNAs (lncRNA) are a novel class of transcripts with no protein coding capacity, but with diverse functions in cancer cell proliferation, apoptosis, and metastasis. The lncRNA PVT1 is 1,716 nt in length and located in the chr8q24.21 region, which also contains the myelocytomatosis (MYC) oncogene. Previous studies demonstrated that MYC promotes PVT1 expression in primary human cancers. However, the expression pattern and potential biologic function of PVT1 in non-small cell lung cancer (NSCLC) is still unclear. Here, we found that PVT1 was upregulated in 105 human NSCLC tissues compared with normal samples. High expression of PVT1 was associated with a higher tumor-node-metastasis stage and tumor size, as well as poorer overall survival. Functional analysis revealed that knockdown of PVT1 inhibited NSCLC cell proliferation and induced apoptosis both in vitro and in vivo RNA immunoprecipitation and chromatin immunoprecipitation assays demonstrated that PVT1 recruits EZH2 to the large tumor suppressor kinase 2 (LATS2) promoter and represses LATS2 transcription. Furthermore, ectopic expression of LATS2 increased apoptosis and repressed lung adenocarcinoma cell proliferation by regulating the Mdm2-p53 pathway. Taken together, our findings indicated that PVT1/EZH2/LATS2 interactions might serve as new target for lung adenocarcinoma diagnosis and therapy. Mol Cancer Ther; 15(5); 1082-94. ©2016 AACR.

[Influence of synthetic technology on the structure and luminescent properties of Sr3B2O6:Eu2+ yellow phosphor].

The Sr3B2O6:Eu(2+) yellow phosphor for warm white LED was synthesized by high temperature solid phase method. The influences on the phosphor structure and luminous properties of sintering temperature and holding time were systematically studied. Results indicated that the optimum synthetic temperature and soaking is 1 150 degrees C and 2 hours respectively. The crystalline structure of phosphor is rhombohedral Sr3B2O6. Sintering temperature and holding time has a significant influence on grain development. The excitation spectrum of phosphor composes of a wide-band spectrum main peaking at 398 nm, and the phosphor can be excited by near ultraviolet and blue light. The luminescence spectrum of phosphor is a broad spectrum peaking at 574 nm. Sintering temperature and holding time have a main effect on luminous intensity of phosphor.