RESUMO
PURPOSE: The aim of this study was to assess the test-retest reliability of a smartphone-based hearing test, performed without supervision of a hearing professional in an uncontrolled environment. METHOD: The hearing application is based on an automated hearing test (DuoTone) and relies on verification procedures (ambient noise monitoring algorithm, graphical user interface) to ensure appropriate measurement conditions. Thresholds obtained with DuoTone were compared to those obtained with standard clinical audiometry for 0.5, 1, 2, and 4 kHz in 13 subjects. Subsequently, test-retest reliability was analyzed using anonymized cloud-stored data from a large group of app users (1,641 subjects) who performed multiple hearing tests. Thresholds at minimum or maximum presentation level of the hearing test (10 dB HL, 85 dB HL) were excluded to avoid floor/ceiling effects. A subset (500 subjects) was created to exclude potentially unreliable data. Test-retest thresholds were compared at 12 test frequencies, from 125 Hz to 12 kHz. RESULTS: Thresholds determined by DuoTone and clinical audiometry did not differ significantly for each test frequency. Regarding test-retest analysis, the percentage of test-retest results within 5 dB ranged from 60% to 77% per test frequency. Results from the subset were not substantially different. Test-retest reliability for app users was comparable to results published in the literature regarding test-retest reliability of audiometry, performed in the clinic. CONCLUSIONS: Initial validation results suggest that thresholds obtained with DuoTone are comparable to clinical audiometry (four frequencies tested). The hearing app provides reliable hearing thresholds between 15 and 80 dB HL (12 frequencies tested) with a test-retest reliability comparable to clinical audiometry.
