Automated Pain Detection in Pediatric Care

Unmet Need: Continuous pain assessment in infants

Continuing advances in intensive care have made it possible for premature infants to survive outside the uterus in Neonatal Intensive Care Unit (NICU) even at very early stages of development. Infants in the NICU may receive an estimated 10-14 painful procedures per day for the first couple of weeks postpartum to sustain their lives. Unfortunately, the pain is poorly assessed and poorly treated in this population. Studies have shown that experiencing pain can have detrimental health effects including programmed cell death. Clinical pain measurement rests on the subjective observation and evaluation of each clinician and may introduce variability in judgement and poor reliability. Currently, the pain measurements are point-in-time or valid only for the specific moment that the infant is observed.

The Technology: Automated Pain Detection in Preterm Infants using Highly-Sensitive Sensors and Data Fusion

The proposed technology is a bedside device that continuously monitors three physio-behavioral signals implicated in the pain expression of premature infants: heart rate variability (HRV), finger splaying (FLEX) and facial grimacing or facial expression (FACE). Neural network (NN) and machine learning techniques are used and trained to associate the three infant signals to a composite pain score. The output value is designed to lead to an estimation of the classification of each signal which matches expert human judgement. By providing continuous, objective pain assessment, this system addresses several major shortcomings in current methods for evaluating pain in infants.

Applications:

• NICUs with modern monitoring capabilities.

• Detection of short-term, long-term, and profound effects of pain exposure in prematurity.

• To improve pain treatment in infants and prevent developmental outcomes.

Advantages:

• Real-time assessment of pain in premature infants.

• On-demand, around-the-clock objective pain measurement.

• Capability to detect the presence of spontaneous pain and its implications in underlying pathology.

Patent Information:

Provisional patent application filed on 09/04/2018.