Close

Presentation

AI Scribes and the Future of Healthcare: A New Paradigm for Informed Consent
DescriptionIn the evolving healthcare technology landscape, Ambient Scribe Technology - also called Ambient Artificial Intelligence (AI) Scribe Technology (herein AI scribe) represents a significant innovation that can revolutionize primary care. AI scribes are designed to semi-automate physician paperwork by operating unobtrusively in the background from electronic devices such as laptops or smartphones. Their operation entails recording patient-physician conversations, transcribing the interactions into text, and subsequently generating summaries in a standard format based on medically relevant information extracted from the dialogue. Some AI scribes can also pre-populate prescriptions, requisitions, and billing codes, further assisting physicians in their daily tasks.

The primary objective of an AI scribe is to alleviate the documentation burden on physicians, a significant factor contributing to physician burnout [1], [2]. However, introducing such technology also brings forth complex challenges related to privacy, consent, and data security [3]. These challenges are particularly pronounced in ambient scribe technology due to its ability to record all information in patient-physician encounters, which may include sensitive and personal details. Moreover, the lack of a comprehensive regulatory framework complicates these issues further [4], [5]. Currently, terms of service agreements between AI scribe providers and physicians govern data collection, retention, and usage [6], [7], [8], [9], [10], [11]. Accordingly, clinicians are tasked with obtaining consent from their patients for the collection, use, and disclosure of personal information. There is ambiguity around service providers' data processing methods, including whether and when de-identification is completed.

At the surface level, the practice aligns with relevant regulatory policies such as the Health Insurance Portability and Accountability Act (HIPAA), the General Data Protection Regulation (GDPR), and the Personal Information Protection and Electronic Documents Act (PIPEDA), based on the premise that users (physicians and patients) provided consent. However, the process undermines the need for informed consent (applicable under GDPR and PIPEDA), the requirement for transparency on data use and processing, and the right to access, amend, or remove data.

This study critically examines and addresses the nuanced issues of privacy and consent arising from deploying AI scribes in clinical settings. Here, we present the evaluation of a Multi-Tiered Granular Informed Consent (MTGIC) designed for AI scribes. The MTGIC design incorporates tiered consent options [12] and granular specificity clearly and concisely. Tiers categorize consent options based on functionalities, such as basic use, data storage, and secondary data use and sharing, whereas granular options provide specific choices within the tiers for data type, duration and types of secondary uses. The framework design draws upon existing literature on ambient intelligent systems and is informed by principles of Value-Sensitive Design [13] and Privacy by Design [14].

The study is guided by key questions aimed at exploring the effectiveness of the MTGIC framework in addressing privacy and consent concerns among patients and physicians, its potential to enhance user adoption and trust, its perceived usability, and the integration of the multi-consent process into clinical workflows.
Authors/Contributorses
Professor in Systems Design Engineering and Associate Vice President, Health Initiatives
Event Type
Lecture
TimeWednesday, September 11th11:15am - 11:35am MST
LocationGrand Ballroom
Tracks
Health Care