Predicting Non-Invasive Ventilation Failure in Pediatric ICU
Author Information
Author(s): Chiaruttini Maria Vittoria, Lorenzoni Giulia, Daverio Marco, Marchetto Luca, Izzo Francesca, Chidini Giovanna, Picconi Enzo, Nettuno Claudio, Zanonato Elisa, Sagredini Raffaella, Rossetti Emanuele, Mondardini Maria Cristina, Cecchetti Corrado, Vitale Pasquale, Alaimo Nicola, Colosimo Denise, Sacco Francesco, Genoni Giulia, Perrotta Daniela, Micalizzi Camilla, Moggia Silvia, Chisari Giosuè, Rulli Immacolata, Wolfler Andrea, Amigoni Angela, Gregori Dario
Primary Institution: University of Padova
Hypothesis
This study aims to identify predictors of first-attempt NIV failure in PICU patients by testing various machine learning techniques and comparing their predictive abilities.
Conclusion
The study identified highly sensitive models for predicting NIV failure in PICU patients, with the Random Forest model being a robust option.
Supporting Evidence
- The RF model demonstrated the highest AUROC and sensitivity equal to 0.83.
- Base excess, weight, age, systolic blood pressure, and fraction of inspired oxygen were identified as the most predictive features.
- Out of 43,794 records, 1861 admissions met the inclusion criteria, with 678 complete cases and 97 NIV failures.
Takeaway
Doctors can use a special computer program to help them figure out which kids might have trouble with a breathing machine, so they can help them faster.
Methodology
Data were sourced from the TIPNet registry, and various machine learning techniques were used to develop predictive models for NIV failure.
Limitations
The study lacked specific disease indicators such as respiratory rate and oxygen saturation, which could affect the model's predictive ability.
Participant Demographics
Patients aged 31 days to 16 years admitted to 23 Italian PICUs.
Statistical Information
P-Value
p<0.05
Confidence Interval
95% CI: 0.64, 0.94
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website