Abstract
Infants born pre-term are at an increased risk for developmental, behavioral, and motor delay and subsequent disability. When these problems are detected early, clinical intervention can be effective at improving functional outcomes. Current methods of early clinical assessment are resource intensive, require extensive training, and do not always capture infants' behavior in natural play environments. We developed the Play and Neuro Development Assessment (PANDA) Gym, an affordable, mechatronic, sensor-based play environment that can be used outside clinical settings to capture infant visual and motor behavior. Using a set of classification codes developed from the literature, we analyzed videos from 24 pre-term and full-term infants as they played with each of three robotic toys designed to elicit different types of interactions—a lion, an orangutan, and an elephant. We manually coded for frequency and duration of toy interactions such as kicking, grasping, touching, and gazing. Pre-term infants gazed at the toys with similar frequency as full-term infants, but infants born full-term physically engaged more frequently and for longer durations with the robotic toys than infants born pre-term. While we showed we could detect differences between full-term and pre-term infants, further work is needed to determine whether differences seen were primarily due to age, developmental delays, or a combination.
Introduction
Approximately 1 in 10 infants worldwide are born preterm, a condition that significantly increases the risk for developmental, behavioral, and motor delays. Early detection of these issues is paramount, as clinical intervention during early infancy can lead to improved functional outcomes. However, existing clinical assessment tools often face challenges: they are resource-intensive, requiring specialized training; they often take place in sterile, non-naturalistic environments; and they frequently rely on qualitative, subjective observations.
To address these limitations, researchers developed the Play and Neuro Development Assessment (PANDA) Gym. This system is designed to be an affordable, portable, and objective tool for quantifying infant behavior in naturalistic play settings, potentially bridging the gap between clinical evaluation and home-based monitoring.
The PANDA Gym System
The PANDA Gym is a "mechatronic playground" engineered to capture a wide array of infant interactions. The setup includes a pressurized force-sensing mat to detect movement and posture, and four GoPro cameras for stereo vision to record behavior from multiple angles. Central to the gym are three specialized robotic toys, each designed to elicit specific motor responses:
- The Elephant: Focuses on upper-limb interaction, such as grasping the trunk or touching the ears. Interaction triggers elephant sounds and LED feedback.
- The Orangutan: Designed for bimanual (two-handed) play. Its arms are connected via magnetic reed switches; pulling them apart activates engaging audio feedback.
- The Lion: Positioned near the infant's feet to encourage lower-limb interactions, specifically kicking.
Methodology and Data Analysis
The study involved a pilot group of 24 infants, categorized into full-term (n=14, average age 5.73 months) and pre-term (n=9, average age 6.36 months uncorrected). Researchers utilized a standardized coding scheme to manually analyze video data, tracking the frequency and duration of seven key behaviors: involuntary contact, gaze, mouthing, hand touch, hand grasp, foot touch, and foot kick.
The analysis focused on "voluntary interaction," defined as purposeful engagement with the toys beyond accidental contact. Out of the initial trials, a subset was discarded due to usability issues, leaving a final dataset of 60 coded video trials.
Key Findings: The Engagement Gap
The study revealed a significant "engagement gap" between full-term and pre-term infants, characterized by a dissociation between visual interest and physical action.
Visual Interest vs. Motor Execution
One of the most striking results was that pre-term infants often displayed visual interest (gazing) similar to their full-term peers, yet failed to translate this interest into physical interaction. For instance, with the Elephant toy, pre-term infants had a median of 0 seconds of voluntary interaction despite the two-minute trial duration. In contrast, full-term infants spent a median of 90.7 seconds interacting ($p = 0.04$).
Toy-Specific Engagement
The Orangutan toy, positioned at eye level near the belly, was the most successful at eliciting interaction from both groups. However, even here, the median interaction time for pre-term infants was less than half that of the full-term group (27.9s vs 69.3s). The Lion toy, placed at the feet, was the least engaging overall, suggesting that infants in this developmental stage (3–11 months) are more visually driven by objects within their immediate upper-body workspace.
Discussion and Future Directions
While this pilot study successfully demonstrated that the PANDA Gym can detect behavioral differences between full-term and pre-term infants, several limitations remain. The small sample size (n=23) limits generalizability. Furthermore, because the groups were matched by chronological (uncorrected) age, some observed differences might reflect developmental timing rather than permanent delays.
Future work will focus on:
- Automation: Moving beyond manual video coding to use the gym's integrated sensors (mat and toys) for automatic behavior detection.
- Longitudinal Studies: Determining if these early "interaction signatures" can serve as predictive markers for long-term neurodevelopmental outcomes, such as Cerebral Palsy.
- Home Deployment: Testing the system's efficacy in non-clinical, low-resource settings to improve accessibility to early screening.
Summary updated