A Framework for Addressing the Risks and Opportunities In AI-Supported Virtual Health Coaches

Abstract

Virtual coaching has rapidly evolved into a foundational component of modern clinical practice. At a time when healthcare professionals are in short supply and the demand for low-cost treatments is ever-increasing, virtual health coaches (VHCs) offer intervention-on-demand for those limited by finances or geographic access to care. More recently, AI-powered virtual coaches have become a viable complement to human coaches. However, the push for AI-powered coaching systems raises several important issues for researchers, designers, clinicians, and patients. In this paper, we present a novel framework to guide the design and development of virtual coaching systems. This framework augments a traditional data science pipeline with four key guiding goals: reliability, fairness, engagement, and ethics.

arXiv comment4 pages

References in corpus (4)

• Beyond Parity: Fairness Objectives for Collaborative Filtering
• Cluster-based Approach to Improve Affect Recognition from Passively Sensed Data
• Learning Models from Data with Measurement Error: Tackling Underreporting
• Extracting Domain Invariant Features by Unsupervised Learning for Robust Automatic Speech Recognition