Standardised hearing loss risk profiles with state-space models

Hearing loss is a growing public health concern, affecting millions and contributing to impaired communication, social isolation, and reduced quality of life. As a hidden condition, hearing loss is typically diagnosed through behavioural tests like pure-tone audiometry, which often fail to capture the full extent of auditory deficits. Additional tests, such as speech-in-quiet and speech-in-noise, provide a more detailed understanding, yet they are underutilised due to equipment and time constraints. To explore test complementarity, we developed an advanced method for profiling hearing loss by integrating audiogram and speech test data using state-space models (SSMs). Our approach – the Campi-Peters-Morvan-Buhl-Thai-Van (CPMBT) model – accounts for unobservable aspects of hearing loss and infers common trends across population segments. We established a baseline model using only audiograms and an extended version incorporating speech tests to capture multiple dimensions of auditory function. Through rigorous statistical testing, we demonstrated that the extended model significantly outperforms the baseline model, particularly for individuals with slight to moderate hearing loss. Our analysis revealed distinctive frequency-specific patterns in how speech tests contribute to hearing assessment. The CPMBT model provides standardized risk profiles that quantify age-related and frequency-dependent patterns in hearing loss progression, establishing a national benchmark for clinical practice and public health strategies.