Introduction Athletes and military personnel are both at risk of disabling injuries due to extreme physical activity. A method to predict which individuals might be more susceptible to injury would be valuable, especially in the military where basic recruits may be discharged from service due to injury. We postulate that certain body characteristics may be used to predict risk of injury with physical activity. Methods US Army basic training recruits between the ages of 17 and 21 (N = 17,680, 28% female) were scanned for uniform fitting using the 3D body imaging scanner, Human Solutions of North America at Fort Jackson, SC. From the 3D body imaging scans, a database consisting of 161 anthropometric measurements per basic training recruit was used to predict the probability of discharge from the US Army due to injury. Predictions were made using logistic regression, random forest, and artificial neural network (ANN) models. Model comparison was done using the area under the curve (AUC) of a ROC curve. Results The ANN model outperformed two other models, (ANN, AUC = 0.70, [0.68,0.72], logistic regression AUC = 0.67, [0.62,0.72], random forest AUC = 0.65, [0.61,0.70]). Conclusions Body shape profiles generated from a three-dimensional body scanning imaging in military personnel predicted dischargeable physical injury. The ANN model can be programmed into the scanner to deliver instantaneous predictions of risk, which may provide an opportunity to intervene to prevent injury.

A long-standing question in nutrition and obesity research involves quantifying the relationship between body fat and anthropometry. To date, the mathematical formulation of these relationships has relied on pairing easily obtained anthropometric measurements such as the body mass index (BMI), waist circumference, or hip circumference to body fat. Recent advances in 3D body shape imaging technology provides a new opportunity for quickly and accurately obtaining hundreds of anthropometric measurements within seconds, however, there does not yet exist a large diverse database that pairs these measurements to body fat. Herein, we leverage 3D scanned anthropometry obtained from a population of United States Army basic training recruits to derive four subpopulations of homogenous body shape archetypes using a combined principal components and cluster analysis. While the Army database was large and diverse, it did not have body composition measurements. Therefore, these body shape archetypes were paired to an alternate smaller sample of participants from the Pennington Biomedical Research Center in Baton Rouge, LA that were not only similarly imaged by the same 3D scanning machine, but also had concomitant measures of body composition by dual-energy X-ray absorptiometry body composition. With this enhanced ability to obtain anthropometry through 3D scanning quickly of large populations, our machine learning approach for pairing body shapes from large datasets to smaller datasets that also contain state-of-the-art body composition measurements can be extended to pair other health outcomes to 3D body shape anthropometry.

Condorcet clustering methods have the attractive features of producing clusterings which place similar points in the same cluster and dissimilar points in different clusters as well as not requiring a priori specification of the number of clusters. They have the disadvantages of being combinatorially hard and the method produces only convex clusters. We propose a novel modification to this method, which improves it significantly on both accounts and works particularly well when applied to social network type data sets. Specifically, we reduce the domain of the clustering to be over a Delaunay triangulation, whose size scales as O(n^⌊ m/2 ⌋) where n is the number of records and m is the number of attributes used for the clustering. The triangulation also limits focus to local structure, which allows for non-convex clusterings. We demonstrate its use in comparison to other well-known heuristic methods using several constructed datasets, then use it to cluster real-world datasets.

Surveys have become a ubiquitous tool used by decision makers to determine the opinion of a population. Recent research suggests that executing surveys soliciting written text responses provides the most statistically informative data to decision makers, allowing for more accurate inference concerning the opinion of a large population. Current surveys quantify population sentiment with questions requiring Likert scaled responses, forcing respondents to map their opinion to a discreet and constructed quantitative scale. This mapping from qualitative opinion to quantitative sentiment measure often results in inaccurate data with high variance. We examine the accuracy of leveraging sentiment analysis to infer the opinion of a population concerning topical questions on a digital survey. We design an experiment to conduct this investigation, administering a survey to a population of over 300 individuals, gathering thousands of responses to three questions - each requiring Likert scaled and written text responses. We employ a baseline sentiment analysis algorithm with a domain specific lexicon to quantify written text responses, and subsequently execute a experimental hypothesis test to evaluate the statistical difference between written sentiment and Likert scaled distributions and means. Statistical tests used include the Randomization Test and Two Sample t Test. Our results indicate that surveyors may substitute questions requiring written text for Likert scaled responses, while maintaining similar distributions of results.

Academic institutions often assess the efficacy of courses by surveying students. These surveys are critical in structuring course content and evaluating instruction. Given the critical function of surveys for academic institutions, it is essential that surveys obtain data which is precise and accurate. Currently most institutions construct surveys employing the Likert scale: questions that require students to map their opinion on precise topics to a discrete, quantitative domain. These surveys uniformly weight response data, irrespective of student interest. We argue greater accuracy may be obtained by building Student-Directed Discussion Surveys (SDDSs) — surveys with several open-ended, student-directed questions, requiring free text responses. SDDSs retain precision by employing several Natural Language Processing (NLP) techniques including word frequency and sentiment analysis. We use SDDSs to improve course content and evaluate survey accuracy by comparing the results of an SDDS to a Likert-scaled survey administered to an overlapping population. We find that the results of these two survey techniques diverge when topics become increasingly significant to respondents. These results, in addition to the documented issues with Likert-scaled surveys, lead to the conclusion that SDDSs may provide more informative and insightful results.

BackgroundPosture, obesity, and body shape are well‐established predictors of injury and athletic performance. However, due to the manual burden in collecting numerous anthropometric measures, to date, no large database and systematic data analysis determining body shape effects on performance and injury have gone beyond using standard measurements like BMI, waist circumference and hip circumference.ObjectiveTo use machine learning based Artificial Neural Network (ANN) and decision tree analysis on a large nationally representative database of automatically captured body anthropometrics to predict basic training‐related injury.MethodsOver N=20,896soldiers (28% female) recruited for US Army basic training at Fort Jackson, SC were scanned for uniform fitting using the Human Solutions Kinect based 3D imaging technology. Each subject image consisted of 161 body shape measurements. After removing subjects with missing measurements, the remaining subjects were split into injured (91 recruits) and non‐injured (13,296 recruits) groups. 75% of each group was placed into a training group and the other 25% into a testing group. An ANN and Decision Tree model evaluated using the area under the curve(AUC) was developed to predict severe physical injury occurrence during 10 weeks of US Army basic training. Body length measurement was compared against height to evaluate whether certain body proportions were more at risk for injury.ResultsThe AUC for the ANN was 0.79 and the AUC for the Decision Tree model was 0.70. It was determined that body proportions of shorter legs and longer torso more than doubled the risk of injury during US Army basic trainingConclusionsMachine learning models that leverage data sources like those from the newly emerging 3D body image scanners may be used to predict severe injury during vigorous physical activity. These results can be used to develop personalized prevention strategies reducing injury and retaining individuals within their activity protocols.Support or Funding InformationNoneThis abstract is from the Experimental Biology 2018 Meeting. There is no full text article associated with this abstract published in The FASEB Journal.

Rich sources of obesity-related data arising from sensors, smartphone apps, electronic medical health records and insurance data can bring new insights for understanding, preventing and treating obesity. For such large datasets, machine learning provides sophisticated and elegant tools to describe, classify and predict obesity-related risks and outcomes. Here, we review machine learning methods that predict and/or classify such as linear and logistic regression, artificial neural networks, deep learning and decision tree analysis. We also review methods that describe and characterize data such as cluster analysis, principal component analysis, network science and topological data analysis. We introduce each method with a high-level overview followed by examples of successful applications. The algorithms were then applied to National Health and Nutrition Examination Survey to demonstrate methodology, utility and outcomes. The strengths and limitations of each method were also evaluated. This summary of machine learning algorithms provides a unique overview of the state of data analysis applied specifically to obesity.

Stakeholders from academic institutions across the world employ surveys to assess the quality of their work. With surveys these stakeholders attempt to obtain quantified, structured, and directed data in order to make decisions. Often these stakeholders employ long, directed Likert scaled surveys to gain this information. We propose an alternate construction for academic surveys, where stakeholders provide 1-3 open ended "free text" questions, allowing students to lead the discussion. We call this survey methodology "Student Directed Discussion Surveys" (SDDS). SDDS retain the ability to provide quantified, structured, and directed results by employing Natural Language Processing (NLP). We confirm the accuracy of SDDS in relation to traditional Likert scaled surveys with a permutation test, assessing a negligible statistical difference between SDDS and Likert surveys using real data. We then show the utility of SDDS by employing word frequency and sentiment analysis, providing important unbiased decision making information, which is limited when traditional Likert scaled surveys are administered.

The United States (US) Army has over 66,000 soldiers engaged in contingency operations across the world. Current budgetary constraints and an uncertain global security environment require these operations to be executed as efficiently as possible. Base camps are the secured areas where soldiers live when deployed to contingency operations. Base camps impose a significant financial and tactical burden during contingency operations and sub-optimal soldier quality of life decisions have significantly contributed to costs. Quality of life (QOL) refers to the non-security and non-mission related services that directly sustain the mission effectiveness of soldiers. Current US Army base camp tactics, techniques, and procedures (TTPs) do not sufficiently specify QOL services, and more detailed doctrine should be developed to support combat units executing contingency operations. In this investigation we employ quantitative methods to select decisions that improve QOL and inform doctrine. We leverage a QOL function and resource consumption data developed by US Army Natick Soldier Research, Development and Engineering Center's (Natick Labs) to build a model that improves QOL under the constraints of four fundamental resources: fuel, water, waste water, and solid waste. We employ a mixed integer linear program modeling approach and execute sensitivity analysis to evaluate the strength of our results. Our final model is formulated as a chance constraint optimization to address the uncertainty associated with resource availability in contingency operations. Our results provide QOL decisions that reduce resource consumption while maintaining an equivalent QOL level when compared to current TTPs. The model provides quantitative rigor, informing decision makers of specific base camp design principles for the development of doctrine.