Identifying longitudinal-growth patterns from infancy to childhood: a study comparing multiple clustering techniques

Joint clustering multiple longitudinal features: A comparison of methods and software packages with practical guidance

Clustering longitudinal features is a common goal in medical studies to identify distinct disease developmental trajectories. Compared to clustering a single longitudinal feature, integrating multiple longitudinal features allows additional information to be incorporated into the clustering process, which may reveal co-existing longitudinal patterns and generate deeper biological insight. Despite its increasing importance and popularity, there is limited practical guidance for implementing cluster analysis approaches for multiple longitudinal features and evaluating their comparative performance in medical datasets. In this paper, we provide an overview of several commonly used approaches to clustering multiple longitudinal features, with an emphasis on application and implementation through R software. These methods can be broadly categorized into two categories, namely model-based (including frequentist and Bayesian) approaches and algorithm-based approaches. To evaluate their performance, we compare these approaches using real-life and simulated datasets. These results provide practical guidance to applied researchers who are interested in applying these approaches for clustering multiple longitudinal features. Recommendations for applied researchers and suggestions for future research in this area are also discussed.

New approaches and technical considerations in detecting outlier measurements and trajectories in longitudinal children growth data

BACKGROUND

Growth studies rely on longitudinal measurements, typically represented as trajectories. However, anthropometry is prone to errors that can generate outliers. While various methods are available for detecting outlier measurements, a gold standard has yet to be identified, and there is no established method for outlying trajectories. Thus, outlier types and their effects on growth pattern detection still need to be investigated. This work aimed to assess the performance of six methods at detecting different types of outliers, propose two novel methods for outlier trajectory detection and evaluate how outliers affect growth pattern detection.

METHODS

We included 393 healthy infants from The Applied Research Group for Kids (TARGet Kids!) cohort and 1651 children with severe malnutrition from the co-trimoxazole prophylaxis clinical trial. We injected outliers of three types and six intensities and applied four outlier detection methods for measurements (model-based and World Health Organization cut-offs-based) and two for trajectories. We also assessed growth pattern detection before and after outlier injection using time series clustering and latent class mixed models. Error type, intensity, and population affected method performance.

RESULTS

Model-based outlier detection methods performed best for measurements with precision between 5.72-99.89%, especially for low and moderate error intensities. The clustering-based outlier trajectory method had high precision of 14.93-99.12%. Combining methods improved the detection rate to 21.82% in outlier measurements. Finally, when comparing growth groups with and without outliers, the outliers were shown to alter group membership by 57.9 -79.04%.

CONCLUSIONS

World Health Organization cut-off-based techniques were shown to perform well in few very particular cases (extreme errors of high intensity), while model-based techniques performed well, especially for moderate errors of low intensity. Clustering-based outlier trajectory detection performed exceptionally well across all types and intensities of errors, indicating a potential strategic change in how outliers in growth data are viewed. Finally, the importance of detecting outliers was shown, given its impact on children growth studies, as demonstrated by comparing results of growth group detection.

Latent Classes of Anthropometric Growth in Early Childhood Using Uni- and Multivariate approaches in a South African Birth Cohort

Background

Conventional methods for modelling longitudinal growth data focus on the analysis of mean longitudinal trends or the identification of abnormal growth based on cross-sectional standardized z-scores. Latent Class Mixed Modelling (LCMM) considers the underlying heterogeneity in growth profiles and allows for the identification of groups of subjects that follow similar longitudinal trends.

Methods

LCMM was used to identify underlying latent profiles of growth for univariate responses of standardized height, standardized weight, standardized body mass index and standardized weight-for-length/height measurements and multivariate response of joint standardized height and standardized weight measurements from birth to five years for a sample of 1143 children from a South African birth cohort, the Drakenstein Child Health Study (DCHS). Allocations across latent growth classes were compared to better understand the differences and similarities across the classes identified given different composite measures of height and weight as input.

Results

Four classes of growth within standardized height (n1=516, n2=112, n3=187, n4=321) and standardized weight (n1=263, n2=150, n3=584, n4=142), three latent growth classes within Body Mass Index (BMI) (n1=481, n2=485, n3=149) and Weight for length/height (WFH) (n1=321, n2=710, n3=84) and five latent growth classes within the multivariate response of standardized height and standardized weight (n1=318, n2=205, n3=75, n4=296, n5=242) were identified, each with distinct trajectories over childhood. A strong association was found between various growth classes and abnormal growth features such as rapid weight gain, stunting, underweight and overweight.

Conclusions

With the identification of these classes, a better understanding of distinct childhood growth trajectories and their predictors may be gained, informing interventions to promote optimal childhood growth.

A novel systematic pipeline for increased predictability and explainability of growth patterns in children using trajectory features

OBJECTIVE

Longitudinal patterns of growth in early childhood are associated with health conditions throughout life. Knowledge of such patterns and the ability to predict them can lead to better prevention and improved health promotion in adulthood. However, growth analyses are characterized by significant variability, and pattern detection is affected by the method applied. Moreover, pattern labelling is typically performed based on ad hoc methods, such as visualizations or clinical experience. Here, we propose a novel pipeline using features extracted from growth trajectories using mathematical, statistical and machine-learning approaches to predict growth patterns and label them in a systematic and unequivocal manner.

METHODS

We extracted mathematical and clinical features from 9577 children growth trajectories embedded with machine-learning predictions of the growth patterns. We experimented with two sets of features (CAnonical Time-series Characteristics and trajectory features specific to growth), developmental periods and six machine-learning classifiers. Clinical experts provided labels for the detected patterns and decision rules were created to associate the features with the labelled patterns. The predictive capacity of the extracted features was validated on two heterogenous populations (The Applied Research Group for Kids and the 2004 Pelotas Birth Cohort, based in Canada and Brazil, respectively).

RESULTS

Features predictive ability measured by accuracy and F1 score was ≥ 80% and ≥ 0.76 respectively in both cohorts. A small number of features (n = 74) was sufficient to distinguish between growth patterns in both cohorts. Slope, intercept of the trajectory, age at peak value, start value and change of the growth measure were among the top identified features.

CONCLUSION

Growth features can be reliably used as predictors of growth patterns and provide an unbiased understanding of growth patterns. They can be used as tool to reduce the effort to repeat analysis and variability concerning anthropometric measures, time points and analytical methods, in the context of the same or similar populations.

BMI growth trajectory from birth to 5 years and its sex-specific association with prepregnant BMI and gestational weight gain

Objective

The purpose of the study was to identify the latent body mass index (BMI) z-score trajectories of children from birth to 5 years of age and evaluate their sex-specific association with prepregnant BMI and gestational weight gain (GWG).

Methods

This was a retrospective longitudinal cohort study performed in China. In total, three distinct BMI-z trajectories from birth to 5 years of age were determined for both genders using the latent class growth modeling. The logistic regression model was used to assess the associations of maternal prepregnant BMI and GWG with childhood BMI-z growth trajectories.

Results

Excessive GWG increased the risks of children falling into high-BMI-z trajectory relative to adequate GWG (OR = 2.04, 95% CI: 1.29, 3.20) in boys; girls born to mothers with prepregnancy underweight had a higher risk of low-BMI-z trajectory than girls born to mothers with prepregnancy adequate weight (OR = 1.85, 95% CI: 1.22, 2.79).

Conclusion

BMI-z growth trajectories of children from 0 to 5 years of age have population heterogeneity. Prepregnant BMI and GWG are associated with child BMI-z trajectories. It is necessary to monitor weight status before and during pregnancy to promote maternal and child health.

A novel shape-based approach to identify gestational age-adjusted growth patterns from birth to 11 years of age

Child growth patterns assessment is critical to design public health interventions. However, current analytical approaches may overlook population heterogeneity. To overcome this limitation, we developed a growth trajectories clustering pipeline that incorporates a shape-respecting distance, baseline centering (i.e., birth-size normalized trajectories) and Gestational Age (GA)-correction to characterize shape-based child growth patterns. We used data from 3945 children (461 preterm) in the 2004 Pelotas Birth Cohort with at least 3 measurements between birth (included) and 11 years of age. Sex-adjusted weight-, length/height- and body mass index-for-age z-scores were derived at birth, 3 months, and at 1, 2, 4, 6 and 11 years of age (INTERGROWTH-21st and WHO growth standards). Growth trajectories clustering was conducted for each anthropometric index using k-means and a shape-respecting distance, accounting or not for birth size and/or GA-correction. We identified 3 trajectory patterns for each anthropometric index: increasing (High), stable (Middle) and decreasing (Low). Baseline centering resulted in pattern classification that considered early life growth traits. GA-correction increased the intercepts of preterm-born children trajectories, impacting their pattern classification. Incorporating shape-based clustering, baseline centering and GA-correction in growth patterns analysis improves the identification of subgroups meaningful for public health interventions.

Association between gut MIcrobiota, GROWth and Diet in peripubertal children from the TARGet Kids! cohort (The MiGrowD) study: protocol for studying gut microbiota at a community-based primary healthcare setting

INTRODUCTION

The gut microbiota interacts with diet to affect body health throughout the life cycle. Critical periods of growth, such as infancy and puberty, are characterised by microbiota remodelling and changes in dietary habits. While the relationship between gut microbiota and growth in early life has been studied, our understanding of this relationship during puberty remains limited. Here, we describe the MIcrobiota, GROWth and Diet in peripubertal children (The MiGrowD) study, which aims to assess the tripartite growth-gut microbiota-diet relationship at puberty.

METHODS AND ANALYSIS

The MiGrowD study will be a cross-sectional, community-based study involving children 8-12 years participating in the TARGet Kids!

COHORT

TARGet Kids! is a primary healthcare practice-based research network in Canada. Children will be asked to provide a stool sample, complete two non-consecutive 24-hour dietary recalls and a pubertal self-assessment based on Tanner Stages. Anthropometry will also be conducted. The primary outcome is the association between gut microbiota composition and longitudinal growth from birth until entry into the study. Anthropometrics data from birth will be from the data collected prospectively through TARGet Kids!. Body mass index z-scores will be calculated according to WHO. The secondary outcome is the association between gut microbiota, diet and pubertal stage.

ETHICS AND DISSEMINATION

Ethics approval has been obtained by the Hospital for Sick Children and St. Michael's Hospital-Unity Health, and the University of Toronto. Results will be disseminated in the public and academic sector, including participants, TARGet Kids! primary healthcare physicians teams, scientists via participation in the TARGet Kids! science and physician meetings, conferences and publications in peer-reviewed journals. The MiGrowD study results will help researchers understand the relationships underlying growth, gut microbiota and pubertal maturation in children.