Comparing Deep Learning Approaches for Understanding Genotype × Phenotype Interactions in Biomass Sorghum.

We explore the use of deep convolutional neural networks (CNNs) trained on overhead imagery of biomass sorghum to ascertain the relationship between single nucleotide polymorphisms (SNPs), or groups of related SNPs, and the phenotypes they control. We consider both CNNs trained explicitly on the classification task of predicting whether an image shows a plant with a reference or alternate version of various SNPs as well as CNNs trained to create data-driven features based on learning features so that images from the same plot are more similar than images from different plots, and then using the features this network learns for genetic marker classification. We characterize how efficient both approaches are at predicting the presence or absence of a genetic markers, and visualize what parts of the images are most important for those predictions. We find that the data-driven approaches give somewhat higher prediction performance, but have visualizations that are harder to interpret; and we give suggestions of potential future machine learning research and discuss the possibilities of using this approach to uncover unknown genotype × phenotype relationships.

Webcams, Crowdsourcing, and Enhanced Crosswalks: Developing a Novel Method to Analyze Active Transportation.

INTRODUCTION: Active transportation opportunities and infrastructure are an important component of a community's design, livability, and health. Features of the built environment influence active transportation, but objective study of the natural experiment effects of built environment improvements on active transportation is challenging. The purpose of this study was to develop and present a novel method of active transportation research using webcams and crowdsourcing, and to determine if crosswalk enhancement was associated with changes in active transportation rates, including across a variety of weather conditions. METHODS: The 20,529 publicly available webcam images from two street intersections in Washington, DC, USA were used to examine the impact of an improved crosswalk on active transportation. A crowdsource, Amazon Mechanical Turk, annotated image data. Temperature data were collected from the National Oceanic and Atmospheric Administration, and precipitation data were annotated from images by trained research assistants. RESULTS: Summary analyses demonstrated slight, bi-directional differences in the percent of images with pedestrians and bicyclists captured before and after the enhancement of the crosswalks. Chi-square analyses revealed these changes were not significant. In general, pedestrian presence increased in images captured during moderate temperatures compared to images captured during hot or cold temperatures. Chi-square analyses indicated the crosswalk improvement may have encouraged walking and biking in uncomfortable outdoor conditions (P < 0.5). CONCLUSION: The methods employed provide an objective, cost-effective alternative to traditional means of examining the effects of built environment changes on active transportation. The use of webcams to collect active transportation data has applications for community policymakers, planners, and health professionals. Future research will work to validate this method in a variety of settings as well as across different built environment and community policy initiatives.