Predictive Modeling of Brain Age for Evaluating Cognitive Brain Aging in Hearing Loss

1154 

Abstract Submission 

Yen-Jung Pan¹, Chen-Yuan Kuo^1,2, Huei-Yu Tsai¹, Liang-Kung Chen^3,4, Ching-Po Lin^1,4,5

¹Institute of Neuroscience, National Yang Ming Chiao Tung University, Taipei, Taiwan, ²Department of Neurology, Neurological Institute, Taipei Veterans General Hospital, Taipei, Taiwan, ³Taipei Municipal Gan-Dau Hospital (managed by Taipei Veterans General Hospital), Taipei, Taiwan, ⁴Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University, Taipei, Taiwan, ⁵Department of Education and Research, Taipei City Hospital, Taipei, Taiwan

Yen-Jung Pan  
Institute of Neuroscience, National Yang Ming Chiao Tung University
Taipei, Taiwan

Chen-Yuan Kuo  
Institute of Neuroscience, National Yang Ming Chiao Tung University|Department of Neurology, Neurological Institute, Taipei Veterans General Hospital
Taipei, Taiwan|Taipei, Taiwan

Huei-Yu Tsai  
Institute of Neuroscience, National Yang Ming Chiao Tung University
Taipei, Taiwan

Liang-Kung Chen  
Taipei Municipal Gan-Dau Hospital (managed by Taipei Veterans General Hospital)|Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University
Taipei, Taiwan|Taipei, Taiwan

Ching-Po Lin  
Institute of Neuroscience, National Yang Ming Chiao Tung University|Center for Healthy Longevity and Aging Sciences, National Yang Ming Chiao Tung University|Department of Education and Research, Taipei City Hospital
Taipei, Taiwan|Taipei, Taiwan|Taipei, Taiwan

The occurrence of hearing loss in midlife may play a pivotal role in contributing to dementia [1]. Prior research shows that prolonged auditory deprivation not only accelerates whole-brain atrophy and impairs executive function, but also leads to alterations in both auditory and visual connectivity, potentially affecting visual processing abilities [2,3,4]. These findings suggest potential advanced brain aging in hearing loss. However, it remains uncertain whether this phenomenon impacts the whole brain or only accelerates aging in specific cognitions. Recently, the brain age prediction framework, utilizing extensive neuroimaging data and machine learning techniques, has been employed to capture and elucidate the process of brain aging [5]. Hence, our aims were to construct brain age frameworks encompassing both the whole brain and specific cognition, and to explore whether individuals with hearing loss undergo advanced brain aging.

T1-weighted MRI data from 1482 healthy participants (age range: 18-92 years; 681 males, 801 females) were obtained from 5 site for training dataset to construct the brain age estimators. The study comprised normal-hearing (NH) group (n=51, 22M/29F) and group with hearing loss (HL) (n=67, 46M/21F) from the I-Lan Longitudinal Aging Study (ILAS). All participants underwent two follow-up examinations, on average 2.28 years apart. T1-weighted image preprocessing followed a previously described procedure and individual gray matter volume (GMV) and gray matter density (GMD) maps were estimated [5]. The hearing-related cognitive masks, encompassing memory, hearing, executive, and vision, were defined through a meta-analysis utilizing a machine learning framework [6]. Subsequently, we utilized the structural covariance network framework with both the whole-brain and the hearing-related gray matter mask to independently extract 600 features of the GM signature (300 GMV and 300 GMD) for the brain age model construction. RBF-SVR algorithm was employed for constructing a brain age estimator with a nested 5-fold cross-validation scheme on the training dataset. Model performance was assessed using the mean absolute error (MAE) and (R²) between chronological and predicted age. Each optimized brain age estimator was then applied to the HL and NH groups to estimate individual predicted brain age and calculate the brain age gap (BAG), representing the difference between chronological age and predicted brain age. To control for confounding effects, analysis of covariance (ANCOVA) was employed to compare global and hearing-related BAGs between NH and HL groups, with age, age², sex, years of education, and total intracranial volume (TIV) included as nuisance variables. Significance was set at p < 0.05.

The constructed global brain age estimator and hearing-related cognitive brain age estimators demonstrated satisfactory performance in the training dataset (global: MAE = 4.27 years, R² = 0.92; executive: MAE = 5.66 years, R² = 0.86; hearing: MAE = 6.81 years, R² = 0.79; memory: MAE = 6.50 years, R² = 0.82; vision: MAE = 6.41 years, R² = 0.82; Figure.1). Using the global brain age estimator, the HL group exhibited a significantly smaller BAG compared with the NH group in Wave 1 (p = 0.034) but not in Wave 2. In both Wave 1 and Wave 2, the HL group showed larger BAGs than the NH group in memory, hearing, and executive function, and a smaller BAG in vision, with no statistical differences (Table.1).

In summary, we presented a framework for hearing-related brain age estimators to explore different cognitive brain aging in hearing loss individuals. Although our study did not yield statistically significant evidence of a noteworthy increase in the BAG within each cognitive brain age estimator for the hearing loss group, we observed an increasing trend in the BAGs for hearing, memory, and executive function. Cognitive brain age estimator showed promise in elucidating the mechanisms of brain aging associated with hearing loss.

Aging ¹

Classification and Predictive Modeling ²

Anatomical MRI

Aging

Cognition

Hearing

Machine Learning

MRI

STRUCTURAL MRI

Other - Brain age