About

Jonathan A. Mitchell, Ph.D., is an Associate Professor of Pediatrics specializing in Gastroenterology, Hepatology, and Nutrition at the Children's Hospital of Philadelphia. His overarching research goal is to help prevent chronic diseases in later life by focusing on energetic behaviors such as sleep, diet, and physical activity, and their impact on body composition during growth and development. His laboratory conducts multidisciplinary epidemiological studies utilizing sensors to measure locomotor activity, advanced imaging for body composition, detailed nutritional assessments, geospatial methods to analyze neighborhood environments, and DNA collection for gene-behavior interaction analyses. Mitchell is also involved in the Unit for Optimizing Behavioral Interventions, which employs the Multiphase Optimization Strategy framework to design digital interventions aimed at improving energetic behavior profiles in childhood. His work integrates various methods to understand and influence behaviors related to health and development, contributing to the fields of pediatric health, nutrition, and chronobiology.

Research topics

• Computer Science
• Geography
• Physical medicine and rehabilitation
• Environmental health
• Environmental science
• Ecology
• Multimedia
• Medicine

Selected publications

• Meal Timing Patterns and Associations with Fat Mass in Adolescents

  medRxiv · 2026-04-23

  articleOpen accessSenior authorCorresponding

  Background: The timing of energy intake could be important in the development of obesity. However, most observational evidence stems from adults, anthropometric defined obesity outcomes, single meal timing phenotyping, and traditional regression modeling. Objective: We aimed to describe meal timing patterns in adolescents and determine if they associated with fat mass by modeling the median and all other percentiles of the frequency distribution. Methods: We analyzed data from the Sleep and Growth Study 2 (S-Grow2, N=286, 12-13y). Participants completed 3-day 24-hour dietary recalls and time stamped eating occasions were used to define 8 meal timing traits, with aide from self-reported wake and bed timing. Principal component analysis (PCA) identified multi-dimensional meal timing patterns. Fat mass index (FMI) was estimated using dual energy X-ray absorptiometry. Quantile regression assessed if there were associations between meal timing traits and FMI across the entire FMI frequency distribution. Results: The typical first and last eating occasions were 8:00am (40 minutes after waking) and 8:00pm (2.7 hours before sleep), respectively, thus the eating period typically lasted 11.5 hours per day. The typical eating period midpoint was 2:15pm, and the timing when 50% of energy intake was consumed typically occurred at 3:15pm. PCA revealed three meal timing patterns: 1) "Delayed Start, Condensed Eating Period" (43% of variance; shorter eating period and delayed timing of first eating); 2) "Late, Sleep Proximal Eating" (30% of variance; later timing of last eating and extended eating period), and 3) "Later Energy Intake" (10% of variance; delayed energy intake midpoint). Higher scores for the "Delayed Start, Condensed Eating Period" pattern associated with higher body mass index and FMI at the upper tails of their distributions. Conclusions: Distinct multidimensional meal timing patterns emerged in early adolescence, with the delayed start, condensed eating period pattern potentially associated with higher adiposity.

  PublisherOA PDFDOI

• Physical Activity Traits From Wrist Sensors Correlate With Clinical Status in Pediatric Pulmonary Hypertension

  Pulmonary Circulation · 2026-01-01

  articleOpen accessSenior author

  ABSTRACT Physical activity (PA) estimated by a wearable sensor may reflect clinical status in pediatric pulmonary hypertension (PH). Prior studies used research‐grade hip‐anchored sensors or commercial wrist sensors with proprietary scoring algorithms. Wrist sensors may offer better acceptability in children; however, their ability to detect associations between PA and clinical characteristics is unknown. Youth 8–18 years with PH [Groups 1–4, functional class (FC) I‐II] and healthy controls wore a GENEActiv accelerometer on the non‐dominant wrist for 14 days. Raw acceleration data were processed using the open‐source GGIR R‐package. Participants completed a 6‐min walk distance (6MWD) and quality‐of‐life questionnaire. Muscle mass and strength were assessed by densitometry and handgrip dynamometry. The most recent cardiac testing was extracted from the medical record. Groups were compared by Fisher's exact test, unpaired t ‐test, or Wilcoxon rank sum test. Multivariate regression models assessed for associations between PA and clinical metrics. Thirty PH participants (median 13.9 years, 57% female, 57% Group 1, 50% FC I) and 29 controls were included. Total PA was similar. PH participants demonstrated fewer and shorter bouts of moderate‐to‐vigorous PA ≥ 10 min and more time spent at lower PA intensities. In PH participants, muscle mass was positively associated with PA but 6MWD was negatively associated with PA. PA was not associated with quality of life. Within the PH group, worse PA traits were associated with lower FC and worse clinical testing. Wrist sensors reveal deficits in PA traits including reduced moderate‐to‐vigorous activity bouts and lower intensity gradients in pediatric PH.

  PublisherDOI

• Association of Stress and Neighborhood Social Context With Actigraphy-Measured and Self-Reported Adolescent Sleep Outcomes

  Journal of Adolescent Health · 2026-05-01

  articleOpen access

  PURPOSE: Chronic stress and unfavorable neighborhood environments may increase adolescents' risk for poor sleep. Few studies have examined whether neighborhood social environments moderate associations of adolescents' perceptions of stress with actigraphy-measured and self-reported sleep outcomes. METHODS: In a cross-sectional study of adolescents aged 15-18 (n = 163) years, perceived stress (10-item Perceived Stress Scale) and perceived neighborhood collective efficacy and safety were assessed via survey. Over 14 days, actigraphy measured nightly sleep duration and timing, while ecological momentary assessment (EMA) measured daily stress, self-reported sleep problems, and sleep environment disruptions. Multivariable mixed effects regression estimated associations of stress and neighborhood social factors with nightly sleep patterns and self-reported sleep outcomes, while multivariable linear regression estimated associations with sleep variability (intraindividual standard deviation of each sleep measure). RESULTS: Higher 10-item Perceived Stress Scale scores and daily EMA-reported stress were associated with more variable sleep duration, onset, and offset timing, and higher and more variable sleep problem scores. Daily EMA-reported stress was also associated with earlier sleep timing. Higher neighborhood collective efficacy was associated with less variable sleep duration and timing, and lower and less variable sleep problem scores. Neighborhood collective efficacy and safety moderated associations between stress and several sleep outcomes (e.g., stronger associations between stress and sleep variability were present among adolescents with low neighborhood safety or collective efficacy). DISCUSSION: Lower perceived stress and higher neighborhood collective efficacy were associated with less variable sleep patterns and lower sleep problem scores. Results suggest positive neighborhood social environments may moderate the relationship of stress with adolescent sleep variability.

  PublisherOA PDFDOI

• Changes in Adolescent Diet Quality During the Middle to High School Transition

  Journal of Nutrition Education and Behavior · 2025-12-05

  articleOpen accessSenior author
  PublisherOA PDFDOI

• Physical activity traits from wrist sensors correlate with clinical status in pediatric pulmonary hypertension

  medRxiv · 2025-10-28

  preprintOpen accessSenior author

  Physical activity (PA) estimated by a wearable sensor may reflect clinical status in pediatric pulmonary hypertension (PH). Prior studies used research-grade hip-anchored sensors or commercial wrist sensors with proprietary scoring algorithms. Wrist sensors offer better acceptability in children, however, their ability to detect associations between PA and clinical characteristics is unknown. Youth 8-18 years with PH [Groups 1-4, functional class (FC) I-II] and healthy controls wore a GENEActiv accelerometer on the non-dominant wrist for 14 days. Raw acceleration data were processed using the open-source GGIR R-package. Participants completed a 6-minute walk distance (6MWD) and quality-of-life questionnaire. Muscle mass and strength were assessed by densitometry and handgrip dynamometry. Most recent cardiac testing was extracted from the medical record. Groups were compared by Fisher's exact test, unpaired t-test, or Wilcoxon rank sum test. Multivariate regression models assessed for associations between PA and clinical metrics. Thirty PH participants (median 13.9 years, 57% female, 57% Group 1, 50% FC I) and 29 controls were included. Total PA was similar. PH participants demonstrated fewer and shorter bouts of moderate-to-vigorous PA ≥10 minutes and more time spent at lower PA intensities. In PH participants, muscle mass was positively associated with PA but 6MWD was negatively associated with PA. PA was not associated with quality-of-life. Within the PH group, worse PA traits were associated with lower FC and worse clinical testing. Wrist sensors reveal deficits in PA traits including reduced moderate-to-vigorous activity bouts and lower intensity gradients in pediatric PH.

  PublisherOA PDFDOI

• The gSOS Polygenic Score is Associated with Bone Density and Fracture Risk in Childhood

  medRxiv · 2025-04-23

  preprintOpen access1st authorCorresponding

  Abstract The polygenic risk score genetic quantitative ultrasound speed of sound (gSOS) was developed using machine learning algorithms in adults of European ancestry and associates with reduced odds of fracture in adults. We aimed to determine if gSOS was associated with bone health in children. Two observational studies of children were evaluated: (1) children enrolled in the Bone Mineral Density in Childhood Study (BMDCS) with genetic data (N=1,727); and (2) children with genetic data for research at the Children’s Hospital of Philadelphia (CHOP; N=10,301). Genetic variants were used to calculate gSOS and genetic ancestry. For the BMDCS, puberty stage, dietary calcium, physical activity and fracture accumulation (none or ≥1 fracture) were self-reported, height and weight were measured and BMI calculated. Areal bone mineral density (aBMD) of the lumbar spine, hip, radius, and whole body were assessed by dual energy X-ray absorptiometry and expressed as Z-scores. The CHOP study paired genetic data with documentation of fracture in the electronic health record (EHR). gSOS associated with higher aBMD Z-scores across 7 skeletal sites [e.g., a 1 SD increase in gSOS associated with 0.17 (95% CI: 0.10-0.24) higher lumbar spine aBMD Z-score]. These associations were consistent for males and females, age, puberty stage, and lifestyle factors, and most consistent among children of European genetic ancestry. A 1 SD increase in gSOS associated with 12% and 16% reduced likelihood of self-reported fracture in the BMDCS (OR=0.84, 95% CI: 0.74, 0.95) and a recorded fracture in the CHOP EHR (OR=0.88; 95% CI: 0.82, 0.95). No sex or genetic ancestry differences were found. A higher gSOS score associated with higher aBMD at multiple skeletal sites and reduced odds of fracture in two independent pediatric samples. This genetic tool may have clinical utility to help enhance bone health in early life and protect against fracture across the lifespan. Lay summary In adults, the polygenic risk score gSOS associates with reduced fracture risk. This study evaluates the relationship of gSOS to bone density and fractures in two groups of children. We found that a 1 standard deviation increase in gSOS was associated with higher bone density at multiple skeletal sites. In our two groups of children, a 1 standard deviation increase in gSOS associated with reduced odds of fracture in children by 12% (95% CI: 0.82, 0.95) and 26% (95% CI: 0.74, 0.95). Having a higher gSOS may enhance bone accretion in early life, and protect against fracture across the lifespan.

  PublisherOA PDFDOI

• Race-neutral Pediatric Reference Ranges for Bone Mineral Density Predict Prospective Fractures in Childhood

  The Journal of Clinical Endocrinology & Metabolism · 2025-03-23 · 6 citations

  article

  INTRODUCTION: Race-specific reference ranges for pediatric areal bone mineral density (BMD) are widely used, but the value of race-based clinical algorithms has been questioned. We developed race-neutral pediatric reference ranges for areal BMD and bone mineral apparent density (BMAD) and compared race-specific vs race-neutral Z-scores in their ability to predict prospective fractures. MATERIAL AND METHODS: This secondary analysis of the Bone Mineral Density in Childhood Study used longitudinal BMD data of the spine, hip, forearm, and total body less head and BMAD from dual-energy x-ray absorptiometry (DXA) scans. Race/ethnicity, dietary calcium, physical activity, and prospective fractures were assessed by questionnaire. Race-neutral reference ranges and height-for-age Z-score adjustment equations were created using the lambda-sigma-mu method. Race-neutral and race-specific Z-scores were compared using linear mixed-effect modeling. Cox proportional hazard modeling was used to test whether race-neutral Z-scores associated with fracture. RESULTS: Race-neutral BMD and BMAD Z-scores were 0.5 to 0.7 SD greater than race-specific Z-scores for Black children but only ∼0.1 SD lower for children from other race/ethnicity groups. Growth and lifestyle factors modified group differences. One SD increase in race-neutral Z-scores was associated with a 12% to 18% reduced risk of fracture. CONCLUSION: We present the first race-neutral pediatric reference ranges for BMD and BMAD that are weighted to be representative of the US population and demonstrate that these Z-scores associate with fracture risk. Adoption of these new reference ranges should be considered, with thoughtful implementation for patients previously monitored with race-specific reference ranges, especially among children who identify as Black.

  PublisherDOI

• Performance of an automated sleep scoring approach for actigraphy data in children and adolescents

  SLEEP · 2025-09-18 · 2 citations

  articleOpen accessSenior author

  STUDY OBJECTIVES: GGIR is an R package for processing raw acceleration data to estimate sleep health parameters. We aimed to (1) assess the performance of three sleep algorithms within GGIR against PSG for detecting sleep/wake in clinically referred, typically-developing children (criterion validity); and (2) describe GGIR-derived sleep estimates from typically developing children enrolled in multiple cohort studies (face validity). METHODS: For criterion evaluation, children (8-16 years, N = 30) wore an actigraphy device for one night during in-lab polysomnography with performance assessed using epoch-by-epoch analyses. For face validity evaluation, four community/free living datasets were used: (1) Bone Mineral Accretion in Young Children (3-5 years, N = 310), (2) School Summer Sleep (5-8 years, N = 118), (3) Sleep and Growth Study 2 (12-13 years; N = 291), and (4) Early Life Exposure to Environmental Toxicants (9-18 years; N = 543). All raw acceleration data were processed using GGIR (v.3.0-0) with the Cole-Kripke (CK), Sadeh (S), and van Hees (vH) algorithm settings. RESULTS: Following the in-lab test, 60 per cent of children were diagnosed with mild to severe obstructive sleep apnea (OSA). For criterion evaluation, the 30-s epoch-by-epoch analyses revealed that average balanced accuracies were 0.80 (Sensitivity = 0.80; Specificity = 0.79), 0.76 (Sensitivity = 0.86; Specificity = 0.65), and 0.67 (Sensitivity = 0.95, Specificity = 0.39) for GGIR-CK, GGIR-vH, and GGIR-S, respectively. For face validity evaluation, sleep estimates mirrored the in-lab performance metrics (e.g. sleep duration estimates were similar when using GGIR-CK and GGIR-VH but approximately 1 h longer when using GGIR-S). CONCLUSIONS: The in-lab performance metrics from typically developing children with and without OSA and cohort-based descriptive statistics from samples of typically developing children provide benchmark data to guide investigators on the suitability of GGIR for automated processing of raw acceleration data for pediatric sleep estimation.

  PublisherOA PDFDOI

• Role of Sleep Duration and Timing on Paediatric <scp>BMI</scp> Across Childhood and Adolescence: Do Both Matter?

  Pediatric Obesity · 2025-10-21 · 1 citations

  articleOpen access

  BACKGROUND: Adequate sleep duration is a prevention factor for paediatric obesity, but the role of timing is still unclear. OBJECTIVES: To investigate associations of sleep duration and timing with BMI in a large paediatric database. METHODS: Medical chart and survey data were collected from 29 409 children aged 2-18 years who attended well-child visits between Jan 2019 and Dec 2023 (repeated-measures cross-sectional design). Logistic and linear mixed effects regression models accounting for repeated measures estimated odds of overweight/obesity and continuous BMI-for-age CDC-based percentiles for each additional/later hour of sleep duration, midpoint (median of bedtime and wake time), and bedtime, adjusted for potential confounders and stratified by age groups. RESULTS: Among young children (2-5 years), shorter sleep duration but not sleep timing was related to higher odds of overweight/obesity (21% higher odds with 95% CI: 3% to 36%). In mid-childhood (6-12 years), shorter sleep duration and later midpoint were associated with higher odds of overweight/obesity (18%, 95% CI = 9%, 26%; 32%, 95% CI = 17%, 49%). Among adolescents (13-18 years), each hour of later sleep midpoint equated to 12% higher odds of living with overweight/obesity (95% CI: 1% to 23%). Linear models were similar. CONCLUSIONS: Shorter sleep duration at younger ages and later sleep timing in adolescence were associated with higher BMI.

  PublisherOA PDFDOI

• 118-OR: Social Determinants of Physical Activity in Young Adults with and without Prediabetes or Type 2 Diabetes

  Diabetes · 2025-06-13

  article

  Introduction and Objective: Prediabetes (PreD) and type 2 diabetes (T2D) are increasingly common in young adulthood, a period of declining physical activity (PA). We investigated associations between individual- and community-level social determinants of health (SDOH) and PA in young adults with and without PreD or T2D. Methods: We conducted a cross-sectional study using survey and Fitbit data from 18-35y young adults in NIH’s All of Us Research Program (data collected 2018-2022). Daily steps surrounding SDOH survey (±14d) were averaged (SDOH by domain in Table). Linear regression adjusting for age, sex, race and PreD/T2D assessed the relationship between steps and SDOH in separate models. Results: The cohort (n=862; age 29.6±3.9y; 78% F; 79% White, 2% Black, 4% Hispanic, 5% Asian, 9% Other; 7% PreD/T2D) averaged 7505±3343 steps/day. Young adults with PreD/T2D had roughly 1000 fewer steps/day. Favorable SDOH that were associated with higher step count included employment, low healthcare discrimination, and greater neighborhood social cohesion (Table). Income, education, insurance, social support, loneliness, and neighborhood disorder were not associated with step count. Conclusion: SDOH varied in their association with PA, which was lower in the setting of PreD/T2D. SDOH-based interventions to increase PA in young adults, particularly those with PreD/T2D, should be investigated. Disclosure M. Jankowski: None. C. Harrison: None. S. Apte: None. J. Mitchell: None. X. Qin: None. M. Vajravelu: None. Funding The Pittsburgh Foundation

  PublisherDOI

Recent grants

• NIH Grant F32CA162847

  NIH · $173k · 2014

• Sleep, Energetic Behaviors, and Adolescent Obesity during the Transition from Middle to High School

  NIH · $787k · 2015–2021

Frequent coauthors

• Babette S. Zemel

  University of Pennsylvania

  127 shared

• Peter James

  106 shared

• Struan F.A. Grant

  97 shared

• J. Aaron Hipp

  North Carolina State University

  95 shared

• Andrea Kelly

  Children's Hospital of Philadelphia

  91 shared

• Jacqueline Kerr

  88 shared

• Suneeta Godbole

  University of Colorado Anschutz Medical Campus

  85 shared

• Shana E. McCormack

  Children's Hospital of Philadelphia

  76 shared