Ethical Considerations for AI-Enabled Voice Analysis: Features of an Emerging Bioethics Literature

With the emergence of voice as a digital biomarker for various health conditions over the last decade, healthcare institutions are exploring avenues to leverage this technology to improve patient outcomes and clinical workflow efficiencies. Voice analysis technologies have the potential to streamline clinical care from speech-based diagnoses to assessing disease risk. This interest in healthcare innovation must also correspond with an understanding of the scope of ethical concerns and focal points that exist within this emerging literature.

Given the significant discourse outside traditional ethics journals, we conducted a review of the literature focusing on artificial intelligence (AI)-enabled voice analysis technologies for clinical diagnosis and monitoring purposes to inform the practice of AI ethics scholars, consultants, and instructors. Studies were published between 2010 and 2023 in PubMed, Web of Science, Scopus, OVID Embase, or Google Scholar. We evaluated 1871 articles, with 48 articles included in the final review that explore the ethics of clinical applications.

Clinical applications of these technologies centered around using voice data for diagnosis of pulmonary and neurological conditions, mental health screening, provider decision-making assistance, and remote patient monitoring. Key themes related to ethics included: ethical design and validation; data privacy and security; health equity; trust, transparency, and accountability; alterations to the physicia-patient relationship; and current guidelines and policies.

As healthcare systems work to create policies and best practices for voice analysis technologies, further research is necessary to understand the perspectives of stakeholders who will be essential for the adoption of these technologies.