Case Study: GoCheck Kids achieves 91% gaze detection accuracy with Provectus AI/ML infrastructure

GoCheck Kids boosts gaze detection from 25% to 91% with Provectus

GoCheck Kids, a pediatric vision screening company, faced a challenge with its mobile app: its gaze detection model only identified when a child wasn't looking at the camera about 25% of the time. This created a manual review process that consumed significant resources. To address this, GoCheck Kids partnered with vendor Provectus to build a new ML infrastructure.

Provectus built an ML experimentation platform on AWS, which included Amazon SageMaker for training. This solution enabled the GoCheck Kids team to rapidly complete over 100 experiments, tripling the gaze detection accuracy to 91%. The improved model was integrated into the app to provide real-time alerts, which immediately reduced manual review work and improved the screening process for healthcare providers.

GoCheck Kids