Current augmentative communication technology has had limited success conveying the needs of some individuals with minimally verbal autism spectrum disorder (mvASD). Primary caregivers report being able to better understand these individuals' non-traditional utterances than those less familiar with the individual such as teachers and community members. We present an eight-month multi-phase case study for a translational platform, ECHOS, that uses primary caregivers' unique knowledge to enhance communicative and affective exchanges between mvASD individuals and the broader community. Through iterative development and participatory design, we discovered that physiological sensors were impractical for long-term use, on-body audio was content-rich and easily accessible, and a custom in-the-moment labeling app was transformative in obtaining accurate labels from caregivers for machine learning advancements. This paper presents the design methodology, results, and reflections from our case study and provides insights into development with and for the special needs community.