A machine learning-based prediction model for gout in hyperuricemics: a nationwide cohort study
Authors' conclusion
The findings of this cohort study suggest that a machine learning-based prediction model had relatively good performance and high NPV for identifying hyperuricemic participants at risk of developing gout.
Study context
- Population
- Adults ≥18 years with at least two serum urate measurements >6.8 mg/dl (2007–2022), without prior gout diagnosis or gout medications; Clalit Health Insurance database, Israel15,055 (5%) diagnosed with gout
- Participants
- 301,385 adults
- Exposure
- XGBoost vs Compact model
- Outcome
- Diagnosis of gout among hyperuricemic patients
Methodological detailsnationwide, machine learning (XGBoost) prediction model, feature selection
- Nationwide
- Machine learning (XGBoost) prediction model
- Feature selection
Main finding
As reported in the abstract
Prediction of gout diagnosis
| Group | ROC AUC | Precision-recall AUC | NPV |
|---|---|---|---|
| XGBoost model | 0.781 (95% CI 0.78-0.784) | 0.208 (95% CI 0.195-0.22) | — not stated in the abstract |
| Compact model | 0.714 (95% CI 0.706-0.723) | — not stated in the abstract | 95% |
ROC AUC
XGBoost model0.781 (95% CI 0.78-0.784)
Compact model0.714 (95% CI 0.706-0.723)
Precision-recall AUC
XGBoost model0.208 (95% CI 0.195-0.22)
Compact model— not stated in the abstract
NPV
XGBoost model— not stated in the abstract
Compact model95%
— not stated in the abstract
Observational data: associations, not effects.
Original publication
A machine learning-based prediction model for gout in hyperuricemics: a nationwide cohort study. Rheumatology (Oxford). Published 19 June 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →