Using Signal Features of Functional Near-Infrared Spectroscopy for Acute Physiological Score Estimation in ECMO
Hsiao-Huang Chang1,2, Kai-Hsiang Hou3, Ting-Wei Chiang3
1Division of Cardiovascular Surgery, Department of Surgery, Taipei Veterans General Hospital, Taipei 11217, Taiwan.
Bioengineering (Basel, Switzerland)
|January 22, 2024
Summary
Near-infrared spectroscopy (NIRS) monitors microcirculation in patients on extracorporeal membrane oxygenation (ECMO). Machine learning effectively distinguishes disease severity, aiding clinical assessment and patient outcome prediction.
Area of Science:
- Cardiovascular Medicine
- Critical Care Medicine
- Biomedical Engineering
Background:
- Extracorporeal membrane oxygenation (ECMO) provides life support for critical cardiopulmonary failure.
- Assessing peripheral microcirculation is crucial for managing patients on prolonged ECMO.
- Technological advancements have increased ECMO utilization.
Purpose of the Study:
- To evaluate near-infrared spectroscopy (NIRS) for monitoring peripheral microcirculation in ECMO patients.
- To develop a machine learning model for disease severity assessment using NIRS data.
- To explore correlations between microcirculation data and clinical severity scores.
Main Methods:
- Non-invasive NIRS was used to monitor knee-level microcirculation in ECMO patients.
- Machine learning algorithms processed NIRS oxygenation data to classify disease severity.
- Clinical parameters were integrated to enhance the predictive model performance.
Main Results:
- The machine learning model successfully distinguished high and low disease severity in both veno-venous (VV-ECMO) and veno-arterial (VA-ECMO) modes.
- Moderate correlations were observed between NIRS data and APACHE II scores in both ECMO groups.
- NIRS demonstrated potential in assessing improvements in patient condition.
Conclusions:
- NIRS is a promising non-invasive tool for evaluating peripheral microcirculation in ECMO patients.
- Machine learning models incorporating NIRS data can aid in clinical severity diagnosis.
- This approach may facilitate timely interventions and improve patient outcomes on ECMO.
