Applying machine learning approaches for predicting obesity risk using US health administrative claims database
Authors' conclusion
Obesity prevalence is under-reported in claims databases. ML models, with or without explicit obesity, show promise in improving obesity prediction accuracy compared with obesity codes alone. Improved obesity status prediction may assist practitioners and payors to estimate the burden of obesity and investigate the potential unmet needs of current treatments.
Study context
- Population
- People ≥18 years with ≥1 body mass index (BMI) reading in MarketScan Explorys Claims-EMR data (January 2013-December 2019)
- Participants
- 692,119 people
- Exposure
- Claims-based BMI diagnosis codes and machine learning (ML) models vs electronic medical records (EMR)-based BMI
- Outcome
- Sensitivity and PPV of BMI-related diagnosis codes; prediction of obesity status (BMI≥30 kg/m²) by ML models (AUC)
Methodological detailsdiagnostic accuracy and machine-learning prediction, EMR claims-linked database, EMR-based BMI as gold standard
- Diagnostic accuracy and machine-learning prediction
- EMR claims-linked database
- EMR-based BMI as gold standard
Main finding
As reported in the abstract
Prediction of obesity vs EMR BMI · study period
| Group | AUC | Sensitivity | PPV |
|---|---|---|---|
| Diagnosis codes | — not stated in the abstract | 16.8-44.8% | 85.4-89.2% |
| XGBoost | 79.4% | — not stated in the abstract | — not stated in the abstract |
AUC
Diagnosis codes— not stated in the abstract
XGBoost79.4%
Sensitivity
Diagnosis codes16.8-44.8%
XGBoost— not stated in the abstract
PPV
Diagnosis codes85.4-89.2%
XGBoost— not stated in the abstract
— not stated in the abstract
Observational data: associations, not effects.
Further findings
| Outcome | Values |
|---|---|
| XGBoost trained without an obesity diagnosis | XGBoost AUC 74.1% |
Feature importance (XGBoost)
- Number of obesity diagnoses and inpatient obesity diagnoses were the most important predictors
Original publication
Applying machine learning approaches for predicting obesity risk using US health administrative claims database. BMJ Open Diabetes Res Care. Published 26 September 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →