Despite progress in perceptual tasks such as image classification, computers still perform poorly on cognitive tasks such as image description and question answering. Cognition is core to tasks that ...involve not just recognizing, but reasoning about our visual world. However, models used to tackle the rich content in images for cognitive tasks are still being trained using the same datasets designed for perceptual tasks. To achieve success at cognitive tasks, models need to understand the interactions and relationships between objects in an image. When asked “What vehicle is the person riding?”, computers will need to identify the objects in an image as well as the relationships
riding(man, carriage)
and
pulling(horse, carriage)
to answer correctly that “the person is riding a horse-drawn carriage.” In this paper, we present the Visual Genome dataset to enable the modeling of such relationships. We collect dense annotations of objects, attributes, and relationships within each image to learn these models. Specifically, our dataset contains over 108K images where each image has an average of
35
objects,
26
attributes, and
21
pairwise relationships between objects. We canonicalize the objects, attributes, relationships, and noun phrases in region descriptions and questions answer pairs to WordNet synsets. Together, these annotations represent the densest and largest dataset of image descriptions, objects, attributes, relationships, and question answer pairs.
Objectives The aim of this study was to define the genetic basis of arrhythmogenic right ventricular cardiomyopathy (ARVC). Background Arrhythmogenic right ventricular cardiomyopathy, characterized ...by right ventricular fibrofatty replacement and arrhythmias, causes sudden death. Autosomal dominant inheritance, reduced penetrance, and 7 desmosome-encoding causative genes are known. The basis of low penetrance is poorly understood. Methods Arrhythmogenic right ventricular cardiomyopathy probands and family members were enrolled, blood was obtained, lymphoblastoid cell lines were immortalized, deoxyribonucleic acid was extracted, polymerase chain reaction (PCR) amplification of desmosome-encoding genes was performed, PCR products were sequenced, and diseased tissue samples were studied for intercellular junction protein distribution with confocal immunofluorescence microscopy and antibodies against key proteins. Results We identified 21 variants in plakophilin-2 ( PKP2 ) in 38 of 198 probands (19%), including missense, nonsense, splice site, and deletion/insertion mutations. Pedigrees showed wide intra-familial variability (severe early-onset disease to asymptomatic individuals). In 9 of 38 probands, PKP2 variants were identified that were encoded in trans (compound heterozygosity). The 38 probands hosting PKP2 variants were screened for other desmosomal genes mutations; second variants (digenic heterozygosity) were identified in 16 of 38 subjects with PKP2 variants (42%), including desmoplakin ( DSP ) (n = 6), desmoglein-2 ( DSG2 ) (n = 5), plakophilin-4 ( PKP4 ) (n = 1), and desmocollin-2 ( DSC2 ) (n = 1). Heterozygous mutations in non -PKP 2 desmosomal genes occurred in 14 of 198 subjects (7%), including DSP (n = 4), DSG2 (n = 5), DSC2 (n = 3), and junctional plakoglobin ( JUP ) (n = 2). All variants occurred in conserved regions; none was identified in 700 ethnic-matched control subjects. Immunohistochemical analysis demonstrated abnormalities of protein architecture. Conclusions These data suggest that the genetic basis of ARVC includes reduced penetrance with compound and digenic heterozygosity. Disturbed junctional cytoarchitecture in subjects with desmosomal mutations confirms that ARVC is a disease of the desmosome and cell junction.
Abstract
Background
Patients with inflammatory bowel disease are using online community forums (OCFs) to seek emotional support. The impact of OCFs on well-being and their emotional content are ...unknown.
Methods
We used an unsupervised machine learning algorithm to identify the thematic content of 51,591 public, online posts from the Crohn’s & Colitis Foundation Community Forum.
Results
We identified 10,702 (20.8%) posts expressing: gratitude (40%), anxiety/fear (20.8%), empathy (18.2%), anger/frustration (13.4%), hope (13.2%), happiness (10.0%), sadness/depression (5.8%), shame/guilt (2.5%), and/or loneliness (2.5%). A common subtheme was the importance of fostering social support.
Conclusions
High-throughput, machine learning-directed analysis of OCFs may help identify psychosocial impacts of inflammatory bowel disease on patients and their caregivers.
Abstract
Background
Patients with inflammatory bowel disease (IBD) are increasingly turning to online community forums (OCFs) to seek and share information, yet little is known about their userbase ...or content, which may dissuade providers from referring patients to OCFs. OCFs offer insight into patient priorities and their psychosocial well-being. We used topic modeling and sentiment analysis to examine sources of emotional distress and coping among IBD patients and their supporters on the largest IBD OCF, the Crohn’s & Colitis Foundation Community Forum.
Methods
We analyzed 51,591 online posts and demographic data from 28,623 users from the Crohn’s & Colitis Foundation Community Forum generated from 1/22/08 to 11/17/17. We extracted self-reported demographic information and post content. We used ATLAS.ti® inductive analysis software to manually identify key themes, including emotions, in the 100 most popular posts of 2008, 2012, and 2016. Nine major emotion themes were observed and a word list of synonyms for each emotion was created. We used the word lists to extract all posts expressing each emotion theme and used latent Dirichlet allocation to perform thematic content analysis. We performed sentiment analysis using the TextBlog library for Python™. Sentiment is scored from -1, the most negative sentiment, to +1, the most positive sentiment.
Results
Among all registered users on the Crohn’s & Colitis Foundation Community Forum, 10,704 (34.4%) contributed at least one post and a subset of 1,455 (5.1%) users were at the 95th percentile of posting activity. All posts generated 2,541,240 views. Compared to less active users (who contributed at least one post), active users were younger (36.6 vs 38.4, p<0.0001), more likely to be female (78% vs 73.5%, p=0.0006), definitively diagnosed with IBD (92.9% vs 88%, p<0.0001), and diagnosed more recently (5.7 years vs 7.6, p<0.0001) (Table 1). Thematic analysis revealed 10,720 posts expressing at least one of 9 major emotion categories: Gratitude (40.0%), anxiety/fear (20.8%), empathy (18.2%), frustration/anger (13.4%), hope (13.2%), happiness (10.0%), depression/sadness (5.8%), shame/guilt (2.5%), and loneliness (2.5%). We identified the top three topics for each emotion (Figure 1). Posts expressing happiness, gratitude, or hope comprised the majority and were associated with more positive mean sentiment scores (0.33, 0.28, 0.12, respectively) compared to all posts on the forum (0.10, p<0.001).
Conclusion
The majority of posts conveying an emotion on the Crohn’s & Colitis Foundation Community Forum were associated with positive sentiment. Posts expressing happiness, gratitude, or hope tended to convey positive messages on the importance of social support and gratitude towards other users in the online community. OCFs are an outlet for those affected by IBD to seek and provide advice in a supportive environment.
Table 1. Crohn’s & Colitis Foundation Community Forum user demographics from 2008 to 2017
Figure 1. Key emotions expressed on the Crohn’s & Colitis Foundation Community Forum and their top three themes
Microtask crowdsourcing has enabled dataset advances in social science and machine learning, but existing crowdsourcing schemes are too expensive to scale up with the expanding volume of data. To ...scale and widen the applicability of crowdsourcing, we present a technique that produces extremely rapid judgments for binary and categorical labels. Rather than punishing all errors, which causes workers to proceed slowly and deliberately, our technique speeds up workers' judgments to the point where errors are acceptable and even expected. We demonstrate that it is possible to rectify these errors by randomizing task order and modeling response latency. We evaluate our technique on a breadth of common labeling tasks such as image verification, word similarity, sentiment analysis and topic classification. Where prior work typically achieves a 0.25x to 1x speedup over fixed majority vote, our approach often achieves an order of magnitude (10x) speedup.
Despite progress in perceptual tasks such as image classification, computers still perform poorly on cognitive tasks such as image description and question answering. Cognition is core to tasks that ...involve not just recognizing, but reasoning about our visual world. However, models used to tackle the rich content in images for cognitive tasks are still being trained using the same datasets designed for perceptual tasks. To achieve success at cognitive tasks, models need to understand the interactions and relationships between objects in an image. When asked "What vehicle is the person riding?", computers will need to identify the objects in an image as well as the relationships riding(man, carriage) and pulling(horse, carriage) in order to answer correctly that "the person is riding a horse-drawn carriage". In this paper, we present the Visual Genome dataset to enable the modeling of such relationships. We collect dense annotations of objects, attributes, and relationships within each image to learn these models. Specifically, our dataset contains over 100K images where each image has an average of 21 objects, 18 attributes, and 18 pairwise relationships between objects. We canonicalize the objects, attributes, relationships, and noun phrases in region descriptions and questions answer pairs to WordNet synsets. Together, these annotations represent the densest and largest dataset of image descriptions, objects, attributes, relationships, and question answers.
The direct and indirect efferent pathways from striatum ultimately reconverge to influence basal ganglia output nuclei, which in turn regulate behavior via thalamocortical and brainstem motor ...circuits. However, the distinct contributions of these two efferent pathways in shaping basal ganglia output are not well understood. We investigated these processes using selective optogenetic control of the direct and indirect pathways, in combination with single-unit recording in the basal ganglia output nucleus substantia nigra pars reticulata (SNr) in mice. Optogenetic activation of striatal direct and indirect pathway projection neurons produced diverse cellular responses in SNr neurons, with stimulation of each pathway eliciting both excitations and inhibitions. Despite this response heterogeneity, the effectiveness of direct pathway stimulation in producing movement initiation correlated selectively with the subpopulation of inhibited SNr neurons. In contrast, effective indirect pathway-mediated motor suppression was most strongly influenced by excited SNr neurons. Our results support the theory that key basal ganglia output neurons serve as an inhibitory gate over motor output that can be opened or closed by striatal direct and indirect pathways, respectively.
Purpose Some research has suggested that discussion of prognosis can disrupt the patient-physician relationship. This study assessed whether physician discussion of prognosis is associated with ...detrimental changes in measures of the strength of the patient-physician relationship. Methods This was a longitudinal cohort study of 265 adult patients with advanced cancer who visited 38 oncologists within community- and hospital-based cancer clinics in Western New York and Northern California. Prognostic discussion was assessed by coding transcribed audio-recorded visits using the Prognostic and Treatment Choices (PTCC) scale and by patient survey at 3 months after the clinic visit. Changes in the strength of the patient-physician relationship were computed as differences in patient responses to The Human Connection and the Perceived Efficacy in Patient-Physician Interactions scales from baseline to 2 to 7 days and 3 months after the clinic visit. Results Prognostic discussion was not associated with a temporal decline in either measure. Indeed, a one-unit increase in PTCC during the audio-recorded visit was associated with improvement in The Human Connection scale at 2 to 7 days after the visit (parameter estimate, 0.10; 95% CI, -0.02 to 0.23) and 3 months after the visit (parameter estimate, 0.18; 95% CI, 0.02 to 0.35) relative to baseline. Standardized effect sizes (SES) associated with an increase of two standard deviations in the PTCC at each time point were consistent with small beneficial effects (SES, 0.14 95% CI, -0.02 to 0.29 at 2 to 7 days; SES, 0.24 95% CI, 0.02 to 0.45 at 3 months), and lower bounds of CIs indicated that substantial detrimental effects of prognostic discussion were unlikely. Conclusion Prognostic discussion is not intrinsically harmful to the patient-physician relationship and may even strengthen the therapeutic alliance between patients and oncologists.
PDF
