Evaluating the effect of heart and respiratory rate measurement errors on the ability to predict the outcome of high flow nasal cannula therapy: a multi-centre study
Authors' conclusion
In this multi-centre study, the recently introduced TabPFN ML model outperformed currently available methods for predicting the outcome of HFNC therapy even when realistic levels of measurement errors were included in the clinical data on RR and HR.
Study context
- Population
- Patients with acute hypoxemic respiratory failure receiving high flow nasal cannula (HFNC); training on RENOVATE trial (Brazil), external validation in Italy and the US596 training, 241 validation
- Comparison
- TabPFN vs rate oxygenation (ROX) index
- Outcome
- Prediction of HFNC therapy outcome (success vs failure) using measurements within the first 2 h, with simulated RR and HR measurement errors
- Follow-up
- First 2 h of treatment
Methodological detailsmulticentre, machine learning, bootstrapping, Monte Carlo simulation of measurement errors
- Multicentre
- Machine learning
- Bootstrapping
- Monte Carlo simulation of measurement errors
Performance
As reported in the abstract
HFNC outcome prediction, external validation
| Test or model | AUC | Accuracy |
|---|---|---|
| TabPFN | 0.86 [0.82-0.89] | 0.79 [0.73-0.84] |
| ROX index | 0.78 [0.75-0.80] | 0.71 [0.68-0.75] |
AUC
TabPFN0.86 [0.82-0.89]
ROX index0.78 [0.75-0.80]
Accuracy
TabPFN0.79 [0.73-0.84]
ROX index0.71 [0.68-0.75]
Further findings
Adding arterial blood gas measurements
- Further improved performance and robustness of TabPFN, but not of the ROX index
Original publication
Evaluating the effect of heart and respiratory rate measurement errors on the ability to predict the outcome of high flow nasal cannula therapy: a multi-centre study. Crit Care. Published 22 November 2025. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →