Music and voice interventions in the NICU: a novel adaptive personalized auditory care based on biophysical feedback

Manuela Filippa, Elda Fischi-Gómez, João Jorge, Vincent Grimaldi, Mathieu Lemay, Jean‐Philippe Thiran, Hervé Lissek, P. H�ppi2024 · DOI

Abstract

Neonatal Intensive Care Units (NICUs) are extremely noisy and stressful environments, known to be detrimental to the neurological outcomes of preterm infants. In a critical period for sensory development, implementing evidence-based acoustic solutions is essential for mitigating these risks. Music and voice interventions have demonstrated beneficial effects on the brain, behavior, and physiological development of preterm infants. Newborns can distinguish between vocal and instrumental melodies, and music intervention creates musical memories while influencing resting-state connectivity. This paper proposes a novel, closed-loop "auditory care" system that uses real-time biophysical feedback to dynamically modulate an infant's acoustic environment, aiming to replicate the protective "bioecological niche" of the womb.

Introduction: The Acoustic Challenge of the NICU

The modern NICU is an acoustically "toxic" environment for preterm infants. Unlike the womb—a low-frequency, fluid-filled chamber dominated by the rhythmic sounds of the maternal heartbeat and voice—the NICU is characterized by unpredictable, high-frequency alarms, mechanical hums, and sudden impulsive noises.

This atypical sensory environment occurs during a critical window of neurodevelopment involving synaptogenesis and myelination. Chronic stress from NICU noise is linked to altered brain structure, specifically in regions like the amygdala and the salience network, leading to long-term neurodevelopmental deficits. Current interventions are often "one-size-fits-all" and fail to account for the infant's immediate physiological state.

Methodology: Adaptive Personalized Auditory Care

The proposed framework functions as a closed-loop system centered on a "Well-being Score" (WBS) to provide tailored auditory enrichment.

1. Multi-Modal Sensing

The system integrates non-obtrusive sensors to monitor the infant in real-time:

  • Optical Sensors (PPG): Used to extract heart rate (HR), heart rate variability (HRV), and respiratory rate (RR).
  • Video Monitoring: Thermal and RGB cameras track motor activity, including movement smoothness, startles, and facial twitches, via dense optical flow algorithms.
  • Low-density EEG: Deep-learning classifiers identify sleep-wake cycles (SWC) to ensure interventions occur at appropriate times.

2. The Well-being Score (WBS)

Data from these sensors are synthesized into a single index. A "Green Light" WBS indicates a stable state suitable for sound enrichment, while a "Red Light" WBS indicates stress (e.g., tachycardia or tremors), triggering a shift in the acoustic strategy toward noise reduction.

3. Acoustic Modulation and Rendering

The system employs two primary technical solutions:

  • Active Noise Control (ANC): Targeted at reducing low-frequency mechanical noise from pumps and fans within the incubator.
  • Binaural Sound Rendering: Delivering personalized music or maternal voice recordings. To maximize naturalism, the system utilizes individual Head-Related Transfer Functions (HRTFs) derived from the infant's own MRI scans.

Key Insights: Timing and Restoration

A critical insight of this research is that timing is paramount; playing music to a stressed infant or one in deep sleep can be counterproductive. By triggering enrichment only during quiet-awake states, the system respects the infant's natural rhythms. Furthermore, the system attempts to restore the "maternal abdomen filter" by using ANC to strip away harsh external noises while reintroducing the familiar, stabilizing frequencies of the maternal voice.

Quantitative Foundations and Results

While the integrated system is a proposal, it is built upon foundational quantitative results from pilot work:

  • Brain Structure: Infants exposed to structured music interventions showed significantly larger amygdala volumes and improved maturation of white matter tracts, such as the acoustic radiation, compared to standard-of-care groups.
  • Functional Connectivity: fMRI data revealed that music-exposed preemies had increased connectivity in the salience network, which is typically weakened by premature birth.
  • Physiological Stability: Maternal voice interventions were shown to stabilize newborns and increase oxytocin levels, providing a biological buffer against pain and stress.

Conclusion and Future Directions

The proposed adaptive care system represents a shift toward personalized neonatal medicine. Future work involves the hardware challenge of medical-grade, "baby-friendly" sensor integration and the logistical scaling of MRI-derived HRTFs. Importantly, this technology is intended to supplement, not replace, human care such as "kangaroo care" (skin-to-skin contact). The next step is a full-scale clinical implementation to validate whether this closed-loop approach outperforms static interventions in long-term cognitive and developmental outcomes.

Summary updated

Abstract

Neonatal Intensive Care Units are an extremely noisy and stressful environment, known to be detrimental to the neurological outcome of preterm born infants.In a time when the sensory environment is critical for the infants' development, implementing new acoustic solutions relying on evidence-based research is key for offsetting these risks.Music and voice interventions have shown beneficial effects on preterm infants' brain, behavior, and physiological development, increasing our understanding of the effects and mechanisms of acoustic interventions for wellbeing.Newborns can distinguish between vocal and instrumental melodies, and music intervention creates musical memories and influences the resting state of preterm infants.Maternal voice interventions stabilize the new-
BibTeX
@misc{filippa2024music,
  author = {Manuela Filippa and Elda Fischi-Gómez and João Jorge and Vincent Grimaldi and Mathieu Lemay and Jean‐Philippe Thiran and Hervé Lissek and P. H�ppi},
  title = {{Music and voice interventions in the NICU: a novel adaptive personalized auditory care based on biophysical feedback}},
  year = {2024},
  doi = {10.61782/fa.2023.1060},
}
Figure