Background: This study aimed to compare one state-of-the-art deep learning method and four classical machine learning methods for classifying mediastinal lymph node metastasis of non-small cell lung cancer (NSCLC) from [superscript 18]F-FDG PET/CT images. Another objective was to compare the discriminative power of the recently popular PET/CT texture features with the widely used diagnostic features such as tumor size, CT value, SUV, image contrast, and intensity standard deviation. The four classical machine learning methods included random forests, support vector machines, adaptive boosting, and artificial neural network. The deep learning method was the convolutional neural networks (CNN). The five methods were evaluated using 1397 lymph nodes collected from PET/CT images of 168 patients, with corresponding pathology analysis results as gold standard. The comparison was conducted using 10 times 10-fold cross-validation based on the criterion of sensitivity, specificity, accuracy (ACC), and area under the ROC curve (AUC). For each classical method, different input features were compared to select the optimal feature set. Based on the optimal feature set, the classical methods were compared with CNN, as well as with human doctors from our institute.

Results: For the classical methods, the diagnostic features resulted in 81~85% ACC and 0.87~0.92 AUC, which were significantly higher than the results of texture features. CNN’s sensitivity, specificity, ACC, and AUC were 84, 88, 86, and 0.91, respectively. There was no significant difference between the results of CNN and the best classical method. The sensitivity, specificity, and ACC of human doctors were 73, 90, and 82, respectively. All the five machine learning methods had higher sensitivities but lower specificities than human doctors.

Conclusions: The present study shows that the performance of CNN is not significantly different from the best classical methods and human doctors for classifying mediastinal lymph node metastasis of NSCLC from PET/CT images. Because CNN does not need tumor segmentation or feature calculation, it is more convenient and more objective than the classical methods. However, CNN does not make use of the import diagnostic features, which have been proved more discriminative than the texture features for classifying small-sized lymph nodes. Therefore, incorporating the diagnostic features into CNN is a promising direction for future research.

Despite wide applications of high-throughput biotechnologies in cancer research, many biomarkers discovered by exploring large-scale omics data do not provide satisfactory performance when used to predict cancer treatment outcomes. This problem is partly due to the overlooking of functional implications of molecular markers. Here, we present a novel computational method that uses evolutionary conservation as prior knowledge to discover bona fide biomarkers. Evolutionary selection at the molecular level is nature's test on functional consequences of genetic elements. By prioritizing genes that show significant statistical association and high functional impact, our new method reduces the chances of including spurious markers in the predictive model. When applied to predicting therapeutic responses for patients with acute myeloid leukemia and to predicting metastasis for patients with prostate cancers, the new method gave rise to evolution-informed models that enjoyed low complexity and high accuracy. The identified genetic markers also have significant implications in tumor progression and embrace potential drug targets. Because evolutionary conservation can be estimated as a gene-specific, position-specific, or allele-specific parameter on the nucleotide level and on the protein level, this new method can be extended to apply to miscellaneous “omics” data to accelerate biomarker discoveries.

The second iteration of the Autism Brain Imaging Data Exchange (ABIDE II) aims to enhance the scope of brain connectomics research in Autism Spectrum Disorder (ASD). Consistent with the initial ABIDE effort (ABIDE I), that released 1112 datasets in 2012, this new multisite open-data resource is an aggregate of resting state functional magnetic resonance imaging (MRI) and corresponding structural MRI and phenotypic datasets. ABIDE II includes datasets from an additional 487 individuals with ASD and 557 controls previously collected across 16 international institutions. The combination of ABIDE I and ABIDE II provides investigators with 2156 unique cross-sectional datasets allowing selection of samples for discovery and/or replication. This sample size can also facilitate the identification of neurobiological subgroups, as well as preliminary examinations of sex differences in ASD. Additionally, ABIDE II includes a range of psychiatric variables to inform our understanding of the neural correlates of co-occurring psychopathology; 284 diffusion imaging datasets are also included. It is anticipated that these enhancements will contribute to unraveling key sources of ASD heterogeneity.

Accumulating data from genome-wide association studies (GWAS) have provided a collection of novel candidate genes associated with complex diseases, such as atherosclerosis. We identified an atherosclerosis-associated single-nucleotide polymorphism (SNP) located in the intron of the long noncoding RNA (lncRNA) LINC00305 by searching the GWAS database. Although the function of LINC00305 is unknown, we found that LINC00305 expression is enriched in atherosclerotic plaques and monocytes. Overexpression of LINC00305 promoted the expression of inflammation-associated genes in THP-1 cells and reduced the expression of contractile markers in co-cultured human aortic smooth muscle cells (HASMCs). We showed that overexpression of LINC00305 activated nuclear factor-kappa beta (NF-κB) and that inhibition of NF-κB abolished LINC00305-mediated activation of cytokine expression. Mechanistically, LINC00305 interacted with lipocalin-1 interacting membrane receptor (LIMR), enhanced the interaction of LIMR and aryl-hydrocarbon receptor repressor (AHRR), and promoted protein expression as well as nuclear localization of AHRR. Moreover, LINC00305 activated NF-κB exclusively in the presence of LIMR and AHRR. In light of these findings, we propose that LINC00305 promotes monocyte inflammation by facilitating LIMR and AHRR cooperation and the AHRR activation, which eventually activates NF-κB, thereby inducing HASMC phenotype switching.

Humans are able to modulate digit forces as a function of position despite changes in digit placement that might occur from trial to trial or when changing grip type for object manipulation. Although this phenomenon is likely to rely on sensing the position of the digits relative to each other and the object, the underlying mechanisms remain unclear. To address this question, we asked subjects (n = 30) to match perceived vertical distance between the center of pressure (CoP) of the thumb and index finger pads (d_y) of the right hand (“reference” hand) using the same hand (“test” hand). The digits of reference hand were passively placed collinearly (d_y = 0 mm). Subjects were then asked to exert different combinations of normal and tangential digit forces (F_n and F_tan, respectively) using the reference hand and then match the memorized d_y using the test hand. The reference hand exerted F_tan of thumb and index finger in either same or opposite direction. We hypothesized that, when the tangential forces of the digits are produced in opposite directions, matching error (1) would be biased toward the directions of the tangential forces; and (2) would be greater when the remembered relative contact points are matched with negligible digit force production. For the test hand, digit forces were either negligible (0.5–1 N, 0 ± 0.25 N; Experiment 1) or the same as those exerted by the reference hand (Experiment 2).Matching error was biased towards the direction of digit tangential forces: thumb CoP was placed higher than the index finger CoP when thumb and index finger F_tan were directed upward and downward, respectively, and vice versa (p < 0.001). However, matching error was not dependent on whether the reference and test hand exerted similar or different forces. We propose that the expected sensory consequence of motor commands for tangential forces in opposite directions overrides estimation of fingertip position through haptic sensory feedback.

National and state organizations have developed policies calling upon afterschool programs (ASPs, 3–6 pm) to serve a fruit or vegetable (FV) each day for snack, while eliminating foods and beverages high in added-sugars, and to ensure children accumulate a minimum of 30 min/d of moderate-to-vigorous physical activity (MVPA). Few efficacious and cost-effective strategies exist to assist ASP providers in achieving these important public health goals. This paper reports on the design and conceptual framework of Making Healthy Eating and Physical Activity (HEPA) Policy Practice in ASPs, a 3-year group randomized controlled trial testing the effectiveness of strategies designed to improve snacks served and increase MVPA in children attending community-based ASPs. Twenty ASPs, serving over 1800 children (6–12 years) will be enrolled and match-paired based on enrollment size, average daily min/d MVPA, and days/week FV served, with ASPs randomized after baseline data collection to immediate intervention or a 1-year delayed group. The framework employed, STEPs (Strategies To Enhance Practice), focuses on intentional programming of HEPA in each ASPs' daily schedule, and includes a grocery store partnership to reduce price barriers to purchasing FV, professional development training to promote physical activity to develop core physical activity competencies, as well as ongoing technical support/assistance. Primary outcome measures include children's accelerometry-derived MVPA and time spend sedentary while attending an ASP, direct observation of staff HEPA promoting and inhibiting behaviors, types of snacks served, and child consumption of snacks, as well as, cost of snacks via receipts and detailed accounting of intervention delivery costs to estimate cost-effectiveness.

Background: Time series models can play an important role in disease prediction. Incidence data can be used to predict the future occurrence of disease events. Developments in modeling approaches provide an opportunity to compare different time series models for predictive power.

Results: We applied ARIMA and Random Forest time series models to incidence data of outbreaks of highly pathogenic avian influenza (H5N1) in Egypt, available through the online EMPRES-I system. We found that the Random Forest model outperformed the ARIMA model in predictive ability. Furthermore, we found that the Random Forest model is effective for predicting outbreaks of H5N1 in Egypt.

Conclusions: Random Forest time series modeling provides enhanced predictive ability over existing time series models for the prediction of infectious disease outbreaks. This result, along with those showing the concordance between bird and human outbreaks (Rabinowitz et al. 2012), provides a new approach to predicting these dangerous outbreaks in bird populations based on existing, freely available data. Our analysis uncovers the time-series structure of outbreak severity for highly pathogenic avian influenza (H5N1) in Egypt.

Background: Many studies used the older ActiGraph (7164) for physical activity measurement, but this model has been replaced with newer ones (e.g., GT3X+). The assumption that new generation models are more accurate has been questioned, especially for measuring lower intensity levels. The low-frequency extension (LFE) increases the low-intensity sensitivity of newer models, but its comparability with older models is unknown. This study compared step counts and physical activity collected with the 7164 and GT3X + using the Normal Filter and the LFE (GT3X+N and GT3X+LFE, respectively).

Findings: Twenty-five adults wore 2 accelerometer models simultaneously for 3Âdays and were instructed to engage in typical behaviors. Average daily step counts and minutes per day in nonwear, sedentary, light, moderate, and vigorous activity were calculated. Repeated measures ANOVAs with post-hoc pairwise comparisons were used to compare mean values. Means for the GT3X+N and 7164 were significantly different in 4 of the 6 categories (p < .05). The GT3X+N showed 2041 fewer steps per day and more sedentary, less light, and less moderate than the 7164 (+25.6, -31.2, -2.9 mins/day, respectively). The GT3X+LFE showed non-significant differences in 5 of 6 categories but recorded significantly more steps (+3597 steps/day; p < .001) than the 7164.

Conclusion: Studies using the newer ActiGraphs should employ the LFE for greater sensitivity to lower intensity activity and more comparable activity results with studies using the older models. Newer generation ActiGraphs do not produce comparable step counts to the older generation devices with the Normal filter or the LFE.

Background: The discovery of genetic associations is an important factor in the understanding of human illness to derive disease pathways. Identifying multiple interacting genetic mutations associated with disease remains challenging in studying the etiology of complex diseases. And although recently new single nucleotide polymorphisms (SNPs) at genes implicated in immune response, cholesterol/lipid metabolism, and cell membrane processes have been confirmed by genome-wide association studies (GWAS) to be associated with late-onset Alzheimer's disease (LOAD), a percentage of AD heritability continues to be unexplained. We try to find other genetic variants that may influence LOAD risk utilizing data mining methods.

Methods: Two different approaches were devised to select SNPs associated with LOAD in a publicly available GWAS data set consisting of three cohorts. In both approaches, single-locus analysis (logistic regression) was conducted to filter the data with a less conservative p-value than the Bonferroni threshold; this resulted in a subset of SNPs used next in multi-locus analysis (random forest (RF)). In the second approach, we took into account prior biological knowledge, and performed sample stratification and linkage disequilibrium (LD) in addition to logistic regression analysis to preselect loci to input into the RF classifier construction step.

Results: The first approach gave 199 SNPs mostly associated with genes in calcium signaling, cell adhesion, endocytosis, immune response, and synaptic function. These SNPs together with APOE and GAB2 SNPs formed a predictive subset for LOAD status with an average error of 9.8% using 10-fold cross validation (CV) in RF modeling. Nineteen variants in LD with ST5, TRPC1, ATG10, ANO3, NDUFA12, and NISCH respectively, genes linked directly or indirectly with neurobiology, were identified with the second approach. These variants were part of a model that included APOE and GAB2 SNPs to predict LOAD risk which produced a 10-fold CV average error of 17.5% in the classification modeling.

Conclusions: With the two proposed approaches, we identified a large subset of SNPs in genes mostly clustered around specific pathways/functions and a smaller set of SNPs, within or in proximity to five genes not previously reported, that may be relevant for the prediction/understanding of AD.

Background: Little research has explored who responds better to an automated vs. human advisor for health behaviors in general, and for physical activity (PA) promotion in particular. The purpose of this study was to explore baseline factors (i.e., demographics, motivation, interpersonal style, and external resources) that moderate intervention efficacy delivered by either a human or automated advisor.

Methods: Data were from the CHAT Trial, a 12-month randomized controlled trial to increase PA among underactive older adults (full trial N = 218) via a human advisor or automated interactive voice response advisor. Trial results indicated significant increases in PA in both interventions by 12 months that were maintained at 18-months. Regression was used to explore moderation of the two interventions.

Results: Results indicated amotivation (i.e., lack of intent in PA) moderated 12-month PA (d = 0.55, p < 0.01) and private self-consciousness (i.e., tendency to attune to one’s own inner thoughts and emotions) moderated 18-month PA (d = 0.34, p < 0.05) but a variety of other factors (e.g., demographics) did not (p > 0.12).

Conclusions: Results provide preliminary evidence for generating hypotheses about pathways for supporting later clinical decision-making with regard to the use of either human- vs. computer-delivered interventions for PA promotion.