Canonical text (the exact bytes fingerprinted)

TRUVACE RECORD VERSION
record: TRV-2026-2273
version: 1
kind: certified
reason: Certified into the record
timestamp: 2026-10-26T18:35:03.892477Z
status: published
lens: g_space
sector: health
headline: Real-World Identification of Obesity: Development and Internal Validation of a Machine Learning Algorithm
dek: To develop and internally validate machine learning (ML) models to retrospectively identify people with obesity (PwO) (body mass index [BMI] ≥ 30 kg/m2) using administrative claims and electronic health record (EHR) data. This study used US commercial and Medicare Advantage claims and EHR data from the Healthcare Integrated Research Database (HIRD) enriched with clinical, laboratory and social drivers of health data from the American Community Survey and Food Access Research Atlas between 1 January 2022 and 30 S…
gain_title: A machine learning model identified people with obesity in claims and health record data more accurately (F1 0.797) than relying on BMI diagnosis codes alone (F1 0.716).
problem_title: (none)
trace_subject: (none)
gain_reading: A machine learning model identified people with obesity in claims and health record data more accurately (F1 0.797) than relying on BMI diagnosis codes alone (F1 0.716).
gain_evidence: The collapsed categorical model (F1 score, 0.797) performed better than a confusion matrix using BMI diagnosis codes alone (F1 score, 0.716). | The ML model successfully identified PwO and outperformed a confusion matrix approach using BMI diagnostic codes alone
problem_reading: (none)
problem_evidence: (none)
quick_read: Researchers built machine learning models using a decision tree method called XGBoost to find people with a body mass index of 30 or higher in US commercial and Medicare Advantage claims and health record data. The data covered January 2022 to September 2024 and were enriched with social drivers of health information.

Better retrospective identification of obesity could support population-level assessments of disease burden, healthcare use, costs and outcomes. The study reports only internal validation, so how well the model performs in other datasets or settings is not shown.
limitation: 
tag: Evidence-backed gain
key_points: The study analyzed 1,728,010 people, of whom 706,992 (40.9%) had obesity. | Among those with obesity, 21.5% were Class 1, 10.9% Class 2 and 8.5% Class 3. | The best collapsed categorical model reached an F1 score of 0.797 against 0.716 for diagnosis codes alone. | Data came from the Healthcare Integrated Research Database, plus the American Community Survey and Food Access Research Atlas.
rundown: The team used US commercial and Medicare Advantage claims and electronic health record data from the Healthcare Integrated Research Database, covering 1 January 2022 to 30 September 2024. They added clinical, laboratory and social drivers of health data.

Models were built with Extreme Gradient Boosting. A random 30% of individuals and their observed BMIs were held out for testing, while the other 70% were used for training and validation.

Performance was judged by area under the precision-recall curve, F1 score, sensitivity, specificity, positive predictive value and accuracy. The F1 score was the basis for comparing the models with BMI diagnosis codes alone.
sources:
- peer_reviewed | Diabetes, Obesity and Metabolism | https://doi.org/10.1111/dom.71468 | 2026-10-25
prev: 0000000000000000000000000000000000000000000000000000000000000000

sha256

713f3e47ac1db7f2fabf8d7897bdaf301d1ab0c4527e285e099de9d0ffb70236

previous

0000000000000000000000000000000000000000000000000000000000000000