In a recent study published in the Scientific Reports Journal, researchers used a machine-learning (ML)-based approach to assess risk factors and body size during the first 1,000 days (between 2 and 4 years). Obesity in adulthood was predicted by tracking body mass index (BMI) values. of age) of life.
study: Predict infant BMI using the first 1000 days of data. Image credit: NicoElNino/Shutterstock.com
Background
The prevalence of obesity in adults and children is increasing significantly worldwide. Early childhood obesity predicts adult obesity, cardiometabolic risk, and childhood morbidity.
Obesity is difficult to treat once it develops and is more likely to persist. Therefore, obesity prevention is a priority in research, and detection of individuals at increased risk of obesity in adulthood could improve prevention efforts.
Modifiable risk factors included elevated maternal BMI values before pregnancy, weight gain during pregnancy, low socioeconomic status, high neonatal weight, and neighborhood-level variables (such as crime and food availability). included. However, data on the potential for risk estimation by combining variables are limited.
Few existing studies estimate childhood obesity, including factors that increase prenatal and early neonatal obesity risk, but 2-4 year olds have more developmental flexibility and more opportunities to influence health behaviors. Some studies have reported
About research
In the current study, researchers used ML algorithms to identify children at high risk for obesity. This may inform policymaking and strategy development for obesity prevention. They also devised a dynamic predictive BMI tracker used in childhood to identify risk of obesity in adulthood.
The research team used least absolute shrinkage and selection operator (LASSO) regression to retain features with the highest coefficients of association with childhood obesity other than height, weight, and body mass index.
They used support vector regression (SVR) with 5-fold cross-validation to determine the developed an estimation model to estimate the BMI of The team excluded individuals who had no clinical symptoms ≥ 1.0 for the entire period.
The steps involved in model development were raw data acquisition and integration, data preprocessing, feature engineering, training, and tracker validation. The tracker was trained using 80.0% of personal data (training dataset) for all time periods.
Electronic health records (EHRs), birth certificates, and geocoded data were obtained from the Early Obesity Prediction (OPEL) registry from 2004 to 2019. According to the Centers for Disease Control and Prevention, the study result was his BMI based on the participant’s age and gender. and prevention (CDC) recommendations.
result
The OPEL Registry consisted of 149,625 visits to 19,724 individuals aged 0.0 to 48.0 months, of which 10,348 were analyzed, of which 4,204, 4,130, and 2,880 were aged 30.0 to 36.0 months. , 36.0 to 42.0 months, and 42.0 to 48.0 months.
After removing false records, imputing missing values, and scaling exposure variables, 50 variables were selected. Nineteen variables were analyzed after LASSO regression, data augmentation and univariate testing.
The model included the following variables: mean height, BMI, weight from 0.0 to 8.0 months, from 8.0 to 16 months, and from 16 to 24 months. The time difference between the last encounter during the period and the encounter two years ago. Mean age, weight, height, BMI, and weight and height percentiles at 2 years of age. Estimated time difference between the last visit two years ago and the target visit during either period.
The tracker was tested using a validation dataset (20.0% of patients) and showed accurate estimates of pediatric BMI (30.0-36.0 months, 36.0-42.0 months, and 42.0-48.0 months). with an average error of 1.0).
Most variables in the model showed significant correlations with childhood BMI across all estimated ranges. The findings demonstrated that trackers can support clinician- and population-level efforts to prevent obesity in the first few days of life.
Modifiable factors associated with elevated childhood BMI were detected during prenatal and early infancy. This includes maternal risk factors during pregnancy, caesarean delivery, infant weight gain at birth, whether the infant wakes up during the night or needs help falling asleep.
Factors such as the proportion of individuals living in food deserts and Hispanic ethnicity are protected from elevated BMI.
Conclusion
Overall, the results of this study demonstrate that early childhood ML and modifiable risk factors can be used to assess pediatric BMI trajectories in an effort to intervene before the onset of unhealthy obesity to reduce the health burden of obesity. It was something that backed up.
Maternal health, child sleep quality, and socioeconomic factors can influence a child’s weight development during early childhood and beyond.
Unlike existing models that estimate BMI using weight cutoffs at specific time points, the BMI tracker uses three 6-month intervals in the future (i.e., 30.0–36.0 months, 36.0–42.0 months, and 42.0 to 48.0 months) can predict BMI. .
This finding may enable pediatricians to monitor changes in BMI over time.
