AI-Assisted Voice Analysis for Laryngeal Cancer: Primary Care Perspectives on Clinical Implementation
Year: 2026
Type: preprint
Study design: Observational Study
Abstract: Background Voice-based artificial intelligence may offer a noninvasive approach to identifying patients at elevated risk for laryngeal cancer, but successful implementation depends on its acceptability and fit within primary care. This study examined primary care clinicians’ approaches to persistent dysphonia, perceptions of artificial intelligence–assisted voice-based risk stratification, and factors influencing clinical implementation. Methods We conducted a cross-sectional mixed-methods study using semi-structured interviews and an anonymous electronic survey that could be completed independently. Interviews explored dysphonia evaluation, referral practices, technology use, and perceived implementation considerations. Survey items assessed clinical reasoning using two dysphonia vignettes, technology readiness, application acceptability, and perceived barriers. Survey responses were summarized using descriptive statistics; no inferential statistical testing was performed. Interviews were analyzed using reflexive thematic analysis, and qualitative and quantitative findings were subsequently integrated at the aggregate level. Results Fifteen clinicians participated in interviews, and 16 anonymous survey responses were received. In a clinical vignette of a patient with a smoking history who presented for persistent dysphonia, all respondents identified laryngeal or other head and neck malignancy in the differential diagnosis, compared with 12 of 15 (80.0%) for a similar clinical vignette without a smoking history. Thirteen of 16 survey respondents (81.3%) reported a positive or very positive initial impression of the proposed application, and 11 (68.8%) were comfortable or very comfortable integrating it into clinical practice. Interviews revealed that clinicians generally lacked a standardized pathway for escalating persistent dysphonia and viewed artificial intelligence primarily as decision support rather than a replacement for clinical judgment. Acceptance was conditional on demonstrated accuracy, minimal workflow disruption, appropriate clinician oversight, efficient referral pathways, and attention to equity, privacy, and accountability. Conclusions Primary care clinicians demonstrated interest in voice-based artificial intelligence for laryngeal cancer risk stratification, but willingness to adopt the technology depended on more than diagnostic performance. Successful implementation will require integration into existing workflows, preservation of clinical oversight, clear pathways from risk identification to definitive evaluation, and attention to patient trust and equitable access.
Source: Research Square
Authors: Cindy C. Ejindu(Data curation, Formal analysis, Investigation, Visualization, Writing – original draft, Writing – review & editing), Meghana Darla(Data curation, Formal analysis, Investigation, Visualization, Writing – review & editing), Karim Asi(Investigation, Writing – original draft), Anthony Law(Conceptualization, Investigation, Methodology, Supervision, Writing – review & editing)
Institutions: Morehouse School of Medicine, Emory University
Language: English
Topic: Voice and Speech Disorders
Subfield: Speech and Hearing
Field: Health Professions
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7236626975
