Respiratory failure & ventilation
A machine learning algorithm for detecting abnormal patterns in continuous capnography and pulse oximetry monitoring
Authors' conclusion
Our algorithm reliably distinguishes normal from abnormal waveforms.
Study context
- Population
- Patients receiving parenteral opioids with continuous capnography and pulse oximetry monitoring on the general care floorsequences (2,944 abnormal)
- Participants
- 7,858 patients
- Comparison
- Five machine learning (ML) models classifying combined capnography and pulse oximetry waveforms as normal or abnormal
- Primary endpoint
- Fβ score (β = 2) for classifying waveforms as normal or abnormal
Methodological detailsmachine learning model development study, prospective, 80% training / 20% test split, Abnormal segments: 60s before and 30s after an abnormal pattern; normal segments: 90s continuous monitoring, randomly sampled
- Machine learning model development study
- Prospective
- 80% training / 20% test split
- Abnormal segments: 60s before and 30s after an abnormal pattern; normal segments: 90s continuous monitoring, randomly sampled
Primary result
As reported in the abstract
Classification performance, test sequences
| Group | Fβ score (β = 2) | Recall | Precision |
|---|---|---|---|
| XGBoost | 0.94 | 0.98 | 0.83 |
XGBoost
Fβ score (β = 2)0.94
Recall0.98
Precision0.83
Further findings
Inter-rater agreement of nine experts on abnormal segments
- High inter-rater agreement
- > 87%
Original publication
A machine learning algorithm for detecting abnormal patterns in continuous capnography and pulse oximetry monitoring. J Clin Monit Comput. Published 15 April 2024. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →