Interpretation and diagnosis of machine learning models have gained renewed interest in recent years with breakthroughs in new approaches. We present Manifold, a framework that utilizes visual analysis techniques to support interpretation, debugging, and comparison of machine learning models in a more transparent and interactive manner. Conventional techniques usually focus on visualizing the internal logic of a specific model type (i.e., deep neural networks), lacking the ability to extend to a more complex scenario where different model types are integrated. To this end, Manifold is designed as a generic framework that does not rely on or access the internal logic of the model and solely observes the input (i.e., instances or features) and the output (i.e., the predicted result and probability distribution). We describe the workflow of Manifold as an iterative process consisting of three major phases that are commonly involved in the model development and diagnosis process: inspection (hypothesis), explanation (reasoning), and refinement (verification). The visual components supporting these tasks include a scatterplot-based visual summary that overviews the models' outcome and a customizable tabular view that reveals feature discrimination. We demonstrate current applications of the framework on the classification and regression tasks and discuss other potential machine learning use scenarios where Manifold can be applied.

Statins have been implicated in the regulation of cell proliferation, apoptosis and tumor progression in cancer patients and statin use at the time of cancer diagnosis has been reported to be associated with reduced cancer risk and improved survival, irrespective of concomitant anti-cancer therapy. A systematic literature search of relevant databases through May 2015 was conducted to identify studies assessing the prognostic impact of statin use on prognostic outcomes in cancer patients. Literature search identified 95 cohort studies that met the inclusion criteria. A meta-analysis of 55 articles showed that statin use was significantly associated with decreased risk of all-cause mortality (HR 0.70, 95% Cl 0.66 to 0.74) compared with nonusers. The observed pooled estimates were retained for cancer-specific mortality (HR 0.60, 95% Cl 0.47 to 0.77), progression-free survival (HR 0.67, 95% Cl 0.56 to 0.81), recurrence-free survial (HR 0.74, 95% Cl 0.65 to 0.83) and disease-free survival (HR 0.53, 95% Cl 0.40 to 0.72). These associations almost remained consistent across those outcomes when stratified by publication type, tumour location, study design, sample size, initiation of statins, disease stage, research country, follow-up duration or research hospital involved. Subgroup analyses according to initiation of statins showed postdiagnosis statin users (HR 0.65, 95% Cl 0.54 to 0.79) gained significantly more recurrence-free survival benefit than prediagnosis statin users (HR 0.86, 95% Cl 0.77 to 0.96) (p for interaction = 0.018). Statin therapy has potential survival benefit for patients with malignancy. Further large-scale prospective studies emphasising survival outcomes of individual cancer type are strongly encouraged.

BackgroundDue to lack of companionship of parents, compared with non left behind children, left behind children (LBC) suffer from more psychological problems compared with children live with their parents. The aim of this study was to explore the mental health status and the relationship among psychological problems and the related factors of LBC.MethodAdopting delaminating-random-group sampling and using region, county, village (town) as sampling framework, we utilized Demographic Data Recording Form, Adolescent Self-Rating Life Events Check List, Scale of APGAR, Perceived Social Support Scale, Simplified Coping Style Questionnaire, Eysenck Personality Questionnaire, Self-Esteem Scale and Scale of Mental Health for Chinese Middle-school Student to assess 1309 left behind child in junior middle school students’ mental health in Hunan. Statistic description, Structural equation model was adopted to analyze the data.ResultThere was a significant difference in score of psychological problems between LBC and non-LBC(F = 18.224, P<0.000). Life event was the major factor(r = .487) that affected psychological problems (path coefficient, PC = 0.08) directly and affect psychological problems indirectly through affecting passive coping (PC = 0.01)and family functioning(PC = 0.02); family functioning impacted psychological problems indirectly through affecting social support (PC = 4.89) and self-esteem (PC = 0.10); social support (PC = −0.02), passive coping (PC =0.07) and active coping PC = −0.04) affected psychological problems directly. Psychoticism (P) (PC = 0.11), Neuroticism (N) influenced psychological problems of LBC both directly (PC = 0.04) and indirectly through affecting self-esteem (PC: P:-1.87; N: -0.83), while Extraversion/Introversion (E) (PC = 0.21) only impact psychological problems indirectly through self-esteem. Altogether, these variables accounted for 50.2% of total variance of psychological problems (F = 130.470, P = 0.000) for LBC.ConclusionIn this research we proved that LBC have more sever psychological problems than non-LBC. We also identified the direct and indirect influential factors of psychological problems of LBC. The findings had important implications for prevention policies and interventions to promote mental health of LBC.

ObjectivesThis study aimed to investigate anxiety sensitivity (AS) in female Chinese nurses to better understand its characteristics and relationship with nursing stress based on the following hypotheses: (1) experienced nurses have higher AS than newly admitted nurses; and (2) specific nursing stresses are associated with AS after controlling general stress.SettingThe cross-sectional survey was conducted from May 2014 to June 2015 among female nurses at the provincial and primary care levels in Hunan Province, China.ParticipantsAmong 793 nurses who volunteered to participate, 745 returned and completed the questionnaires. Eligible participants are healthy female nurses aged 18–55 years and exempt from a history of psychiatric disorder or severe somatic disease and/or a family history of psychiatric disorder.Primary and secondary outcome measuresAS was assessed by the Anxiety Sensitivity Index-3 (ASI-3). Anxiety symptoms, general stress and nursing stress were measured by the Beck Anxiety Inventory (BAI), the Perceived Stress Scale (PSS-10) and the Nursing Stress Scale (NSS).ResultsThere were significant differences overall and in the three dimensions of AS across nurses of different career stages (all p<0.05). Middle and late career nurses had higher AS than early career nurses (all p<0.05), while no significant difference was found between middle and late career nurses. Conflict with physicians and heavy workload had a significant effect on all aspects of AS, whereas lack of support was related to cognitive AS (all p<0.05).ConclusionsAfter years of exposure to stressful events during nursing, experienced female nurses may become more sensitive to anxiety. Middle career stage might be a critical period for psychological intervention targeting on AS. Hospital administrators should make efforts to reduce nurses' workload and improve their professional status. Meanwhile, more social support and appropriate psychological intervention would be beneficial to nurses with higher AS.