This study aims to develop a predictive model for age-related hearing loss (ARHL) based on multi-source risk factors and artificial intelligence techniques. A retrospective analysis will be conducted on 1,000 cases with 15-year longitudinal clinical data, including audiological assessments and noise exposure history. Machine learning algorithms will be employed to construct a predictive model for hearing loss progression. Additionally, a prospective cohort of 100 community-dwelling elderly individuals will be enrolled. Blood samples will be collected for low-abundance targeted proteomics analysis to screen for biomarkers associated with cognitive impairment. This study will establish an early risk identification tool for ARHL and propose strategies for the screening and prevention of dementia in individuals with hearing impairment, thereby providing evidence-based support for early intervention in auditory and cognitive health in the elderly.

Inclusion Criteria:

1. Age ≥ 60 years;
2. Availability of longitudinal pure-tone audiometry data;
3. Documented history of occupational noise exposure;
4. Complete clinical data (including past medical history and medication history).

Exclusion Criteria:

1. Hearing loss caused by non-age or non-noise factors (e.g., otitis media, otosclerosis, Meniere's disease);
2. Missing clinical data >20%;
3. Concurrent severe mental illness or cognitive impairment (unable to complete audiological assessment).

This study enrolls community-dwelling adults aged ≥60 years from multiple Chinese centers. Inclusion: occupational noise exposure, longitudinal pure-tone audiometry, complete clinical data. Exclusion: non-age/noise hearing loss (e.g., otitis media, otosclerosis, Meniere's disease), missing data >20%, severe mental/cognitive impairment.

The prospective cohort (n=100) recruited from community health centers in North and East China. Inclusion: permanent local residents (≥9 months/year), able to complete assessments, WHO ARHL criteria (PTA≥25 dB HL), written consent. Exclusion: severe psychiatric disorders, major organ failure (NYHA III-IV, eGFR<30), life expectancy <3 years, non-ARHL loss, diagnosed dementia, Parkinson's, stroke with severe sequelae, or other unsuitable conditions.

Prospective participants followed at baseline, 12 months. Among them, 50 ARHL with cognitive impairment (MoCA<26) and 50 with ARHL+normal cognition (MoCA≥26) receive proteomics analysis for biomarker discovery