Data science competition for cross-site individual tree species identification from airborne remote sensing data.

Data on individual tree crowns from remote sensing have the potential to advance forest ecology by providing information about forest composition and structure with a continuous spatial coverage over large spatial extents. Classifying individual trees to their taxonomic species over large regions from remote sensing data is challenging. Methods to classify individual species are often accurate for common species, but perform poorly for less common species and when applied to new sites. We ran a data science competition to help identify effective methods for the task of classification of individual crowns to species identity. The competition included data from three sites to assess each methods' ability to generalize patterns across two sites simultaneously and apply methods to an untrained site. Three different metrics were used to assess and compare model performance. Six teams participated, representing four countries and nine individuals. The highest performing method from a previous competition in 2017 was applied and used as a baseline to understand advancements and changes in successful methods. The best species classification method was based on a two-stage fully connected neural network that significantly outperformed the baseline random forest and gradient boosting ensemble methods. All methods generalized well by showing relatively strong performance on the trained sites (accuracy = 0.46-0.55, macro F1 = 0.09-0.32, cross entropy loss = 2.4-9.2), but generally failed to transfer effectively to the untrained site (accuracy = 0.07-0.32, macro F1 = 0.02-0.18, cross entropy loss = 2.8-16.3). Classification performance was influenced by the number of samples with species labels available for training, with most methods predicting common species at the training sites well (maximum F1 score of 0.86) relative to the uncommon species where none were predicted. Classification errors were most common between species in the same genus and different species that occur in the same habitat. Most methods performed better than the baseline in detecting if a species was not in the training data by predicting an untrained mixed-species class, especially in the untrained site. This work has highlighted that data science competitions can encourage advancement of methods, particularly by bringing in new people from outside the focal discipline, and by providing an open dataset and evaluation criteria from which participants can learn.

Preliminary Outcomes of Combined Treadmill and Overground High-Intensity Interval Training in Ambulatory Chronic Stroke.

PURPOSE: Locomotor high-intensity interval training (HIIT) is a promising intervention for stroke rehabilitation. However, overground translation of treadmill speed gains has been somewhat limited, some important outcomes have not been tested and baseline response predictors are poorly understood. This pilot study aimed to guide future research by assessing preliminary outcomes of combined overground and treadmill HIIT. MATERIALS AND METHODS: Ten participants >6 months post-stroke were assessed before and after a 4-week no-intervention control phase and a 4-week treatment phase involving 12 sessions of overground and treadmill HIIT. RESULTS: Overground and treadmill gait function both improved during the treatment phase relative to the control phase, with overground speed changes averaging 61% of treadmill speed changes (95% CI: 33-89%). Moderate or larger effect sizes were observed for measures of gait performance, balance, fitness, cognition, fatigue, perceived change and brain volume. Participants with baseline comfortable gait speed <0.4 m/s had less absolute improvement in walking capacity but similar proportional and perceived changes. CONCLUSIONS: These findings reinforce the potential of locomotor HIIT research for stroke rehabilitation and provide guidance for more definitive studies. Based on the current results, future locomotor HIIT studies should consider including: (1) both overground and treadmill training; (2) measures of cognition, fatigue and brain volume, to complement typical motor and fitness assessment; and (3) baseline gait speed as a covariate.

Fusion neural networks for plant classification: learning to combine RGB, hyperspectral, and lidar data.

Airborne remote sensing offers unprecedented opportunities to efficiently monitor vegetation, but methods to delineate and classify individual plant species using the collected data are still actively being developed and improved. The Integrating Data science with Trees and Remote Sensing (IDTReeS) plant identification competition openly invited scientists to create and compare individual tree mapping methods. Participants were tasked with training taxon identification algorithms based on two sites, to then transfer their methods to a third unseen site, using field-based plant observations in combination with airborne remote sensing image data products from the National Ecological Observatory Network (NEON). These data were captured by a high resolution digital camera sensitive to red, green, blue (RGB) light, hyperspectral imaging spectrometer spanning the visible to shortwave infrared wavelengths, and lidar systems to capture the spectral and structural properties of vegetation. As participants in the IDTReeS competition, we developed a two-stage deep learning approach to integrate NEON remote sensing data from all three sensors and classify individual plant species and genera. The first stage was a convolutional neural network that generates taxon probabilities from RGB images, and the second stage was a fusion neural network that "learns" how to combine these probabilities with hyperspectral and lidar data. Our two-stage approach leverages the ability of neural networks to flexibly and automatically extract descriptive features from complex image data with high dimensionality. Our method achieved an overall classification accuracy of 0.51 based on the training set, and 0.32 based on the test set which contained data from an unseen site with unknown taxa classes. Although transferability of classification algorithms to unseen sites with unknown species and genus classes proved to be a challenging task, developing methods with openly available NEON data that will be collected in a standardized format for 30 years allows for continual improvements and major gains for members of the computational ecology community. We outline promising directions related to data preparation and processing techniques for further investigation, and provide our code to contribute to open reproducible science efforts.

Functional magnetic resonance brain imaging of imagined walking to study locomotor function after stroke.

OBJECTIVE: Imagined walking has yielded insights into normal locomotor control and could improve understanding of neurologic gait dysfunction. This study evaluated brain activation during imagined walking in chronic stroke. METHODS: Ten persons with stroke and 10 matched controls completed a walking test battery and a magnetic resonance imaging session including imagined walking and knee extension tasks. Brain activations were compared between tasks and groups. Associations between activations and composite gait score were also calculated, while controlling for lesion load. RESULTS: Stroke and worse gait score were each associated with lesser overall brain activation during knee extension but greater overall activation during imagined walking. During imagined walking, the stroke group significantly activated the primary motor cortex lower limb region and cerebellar locomotor region. Better walking function was associated with less activation of these regions and greater activation of medial superior frontal gyrus area 9. CONCLUSIONS: Compared with knee extension, imagined walking was less sensitive to stroke-related deficits in brain activation but better at revealing compensatory changes, some of which could be maladaptive. SIGNIFICANCE: The identified associations for imagined walking suggest potential neural mechanisms of locomotor adaptation after stroke, which could be useful for future intervention development and prognostication.

Locomotor training intensity after stroke: Effects of interval type and mode.

Background and Objectives: High-intensity interval training (HIIT) is a promising strategy for improving gait and fitness after stroke, but optimal parameters remain unknown. We tested the effects of short vs long interval type and over-ground vs treadmill mode on training intensity. Methods: Using a repeated measures design, 10 participants with chronic hemiparesis performed 12 HIIT sessions over 4 weeks, alternating between short and long-interval HIIT sessions. Both protocols included 10 minutes of over-ground HIIT, 20 minutes of treadmill HIIT and another 10 minutes over-ground. Short-interval HIIT involved 30 second bursts at maximum safe speed and 30-60 second rest periods. Long-interval HIIT involved 4-minute bursts at ~90% of peak heart rate (HRpeak) and 3-minute recovery periods at ~70% HRpeak. Results: Compared with long-interval HIIT, short-interval HIIT had significantly faster mean overground speeds (0.75 vs 0.67 m/s) and treadmill speeds (0.90 vs 0.51 m/s), with similar mean treadmill HR (82.9 vs 81.8%HRpeak) and session perceived exertion (16.3 vs 16.3), but lower overground HR (78.4 vs 81.1%HRpeak) and session step counts (1481 vs 1672). For short-interval HIIT, training speeds and HR were significantly higher on the treadmill vs. overground. For long-interval HIIT, the treadmill elicited HR similar to overground training at significantly slower speeds. Conclusions: Both short and long-interval HIIT elicit high intensities but emphasize different dosing parameters. From these preliminary findings and previous studies, we hypothesize that overground and treadmill short-interval HIIT could be optimal for improving gait speed and overground long-interval HIIT could be optimal for improving gait endurance.