Transforming Big Data into AI-ready data for nutrition and obesity research.

OBJECTIVE: Big Data are increasingly used in obesity and nutrition research to gain new insights and derive personalized guidance; however, this data in raw form are often not usable. Substantial preprocessing, which requires machine learning (ML), human judgment, and specialized software, is required to transform Big Data into artificial intelligence (AI)- and ML-ready data. These preprocessing steps are the most complex part of the entire modeling pipeline. Understanding the complexity of these steps by the end user is critical for reducing misunderstanding, faulty interpretation, and erroneous downstream conclusions. METHODS: We reviewed three popular obesity/nutrition Big Data sources: microbiome, metabolomics, and accelerometry. The preprocessing pipelines, specialized software, challenges, and how decisions impact final AI- and ML-ready products were detailed. RESULTS: Opportunities for advances to improve quality control, speed of preprocessing, and intelligent end user consumption were presented. CONCLUSIONS: Big Data have the exciting potential for identifying new modifiable factors that impact obesity research. However, to ensure accurate interpretation of conclusions arising from Big Data, the choices involved in preparing AI- and ML-ready data need to be transparent to investigators and clinicians relying on the conclusions.

Predicting energy intake in adults who are dieting and exercising.

BACKGROUND: When a lifestyle intervention combines caloric restriction and increased physical activity energy expenditure (PAEE), there are two components of energy balance, energy intake (EI) and physical activity energy expenditure (PAEE), that are routinely misreported and expensive to measure. Energy balance models have successfully predicted EI if PAEE is known. Estimating EI from an energy balance model when PAEE is not known remains an open question. OBJECTIVE: The objective was to evaluate the performance of an energy balance differential equation model to predict EI in an intervention that includes both calorie restriction and increases in PAEE. DESIGN: The Antonetti energy balance model that predicts body weight trajectories during weight loss was solved and inverted to estimate EI during weight loss. Using data from a calorie restriction study that included interventions with and without prescribed PAEE, we tested the validity of the Antonetti weight predictions against measured weight and the Antonetti EI model against measured EI using the intake-balance method at 168 days. We then evaluated the predicted EI from the model against measured EI in a study that prescribed both calorie restriction and increased PAEE. RESULTS: Compared with measured body weight at 168 days, the mean (±SD) model error was 1.30 ± 3.58 kg. Compared with measured EI at 168 days, the mean EI (±SD) model error in the intervention that prescribed calorie restriction and did not prescribe increased PAEE, was -84.9 ± 227.4 kcal/d. In the intervention that prescribed calorie restriction combined with increased PAEE, the mean (±SD) EI model error was -155.70 ± 205.70 kcal/d. CONCLUSION: The validity of the newly developed EI model was supported by experimental observations and can be used to determine EI during weight loss.

Characterizing population exposure to coal emissions sources in the United States using the HyADS model.

In anticipation of the expanding appreciation for air quality models in health outcomes studies, we develop and evaluate a reduced-complexity model for pollution transport that intentionally sacrifices some of the sophistication of full-scale chemical transport models in order to support applicability to a wider range of health studies. Specifically, we introduce the HYSPLIT average dispersion model, HyADS, which combines the HYSPLIT trajectory dispersion model with modern advances in parallel computing to estimate ZIP code level exposure to emissions from individual coal-powered electricity generating units in the United States. Importantly, the method is not designed to reproduce ambient concentrations of any particular air pollutant; rather, the primary goal is to characterize each ZIP code's exposure to these coal power plants specifically. We show adequate performance towards this goal against observed annual average air pollutant concentrations (nationwide Pearson correlations of 0.88 and 0.73 with SO 4 2 - and PM2.5, respectively) and coal-combustion impacts simulated with a full-scale chemical transport model and adjusted to observations using a hybrid direct sensitivities approach (correlation of 0.90). We proceed to provide multiple examples of HyADS's single-source applicability, including to show that 22% of the population-weighted coal exposure comes from 30 coal-powered electricity generating units.

Gestational growth trajectories derived from a dynamic fetal-placental scaling law.

Fetal trajectories characterizing growth rates in utero have relied primarily on goodness of fit rather than mechanistic properties exhibited in utero. Here, we use a validated fetal-placental allometric scaling law and a first principles differential equations model of placental volume growth to generate biologically meaningful fetal-placental growth curves. The growth curves form the foundation for understanding healthy versus at-risk fetal growth and for identifying the timing of key events in utero.

Using Classroom Data to Teach Students about Data Cleaning and Testing Assumptions.

This paper discusses the influence that decisions about data cleaning and violations of statistical assumptions can have on drawing valid conclusions to research studies. The datasets provided in this paper were collected as part of a National Science Foundation grant to design online games and associated labs for use in undergraduate and graduate statistics courses that can effectively illustrate issues not always addressed in traditional instruction. Students play the role of a researcher by selecting from a wide variety of independent variables to explain why some students complete games faster than others. Typical project data sets are "messy," with many outliers (usually from some students taking much longer than others) and distributions that do not appear normal. Classroom testing of the games over several semesters has produced evidence of their efficacy in statistics education. The projects tend to be engaging for students and they make the impact of data cleaning and violations of model assumptions more relevant. We discuss the use of one of the games and associated guided lab in introducing students to issues prevalent in real data and the challenges involved in data cleaning and dangers when model assumptions are violated.