A Bradycardia-Based Stress Calculator for the Neonatal Intensive Care Unit: A Multisystem Approach

Area of Science:

• Neonatal care
• Neuroscience
• Biomedical engineering

Background:

• Early life stress in the NICU can lead to adverse health outcomes and neurodevelopmental delays in premature infants.
• Current methods lack a data-driven approach to quantify early-life stress using physiological markers.

Purpose of the Study:

• To develop an automatic stress detector for NICU patients.
• To investigate the relationship between bradycardias, hypoxic events, and perinatal stress.
• To establish a method for quantifying early-life stress based on physiological fingerprints.

Main Methods:

• Recorded EEG, ECG, and SpO2 from 136 NICU patients.
• Assessed stress burden using the Leuven Pain Scale.
• Employed subspace linear discriminant analysis models to detect stress based on bradycardic spells.

Main Results:

• Achieved high classification performance with an area under the curve (AUC) of 0.80-0.96 and kappa scores of 0.41-0.80.
• Identified that stress increases SpO2 desaturations and EEG regularity.
• Observed enhanced interaction between cardiovascular and neurological systems during stress.

Conclusions:

• Stress in NICU infants is quantifiable using physiological data, specifically bradycardic events.
• Stress may exacerbate respiratory abnormalities, potentially impacting neurological and behavioral development.
• The developed stress detector can aid in monitoring and managing stress in premature infants.