Machine learning analysis of cardiovascular risk factors and their associations with hearing loss DOI Creative Commons
Seyed Ali Nabavi, Farima Safari, Ali Faramarzi

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 22, 2025

Hearing loss poses immense burden worldwide and early detection is crucial. The accurate models identify high-risk groups, enabling timely intervention to improve quality of life. subtle changes in hearing often go unnoticed, presenting a challenge for detection. While machine learning shows promise, prior studies have not leveraged cardiovascular risk factors known impact hearing. As outcomes remain challenging characterize associations, we evaluated new approach predict current through using factors. National Health Nutrition Examination Survey (NHANES) 2012–2018 data comprising audiometric tests was utilized. Machine algorithms were trained classify impairment thresholds pure tone average values. Key results showed light gradient boosted performing best classifying mild or greater (> 25 dB HL) with 80.1% accuracy. It also classified > 16 HL 40 thresholds, accuracies exceeding 77% 86% respectively. study found that CatBoost Gradient Boosting performed well test set around 0.79 F1-scores 0.79–0.80. A multi-layer neural network emerged as the top predictor averages, achieving mean absolute error just 3.05 dB. Feature analysis identified age, gender, blood pressure waist circumference key associated Findings offer promising direction clinically applicable tool, personalized prevention strategies, calls prospective validation.

Language: Английский

Machine learning analysis of cardiovascular risk factors and their associations with hearing loss DOI Creative Commons
Seyed Ali Nabavi, Farima Safari, Ali Faramarzi

et al.

Scientific Reports, Journal Year: 2025, Volume and Issue: 15(1)

Published: March 22, 2025

Hearing loss poses immense burden worldwide and early detection is crucial. The accurate models identify high-risk groups, enabling timely intervention to improve quality of life. subtle changes in hearing often go unnoticed, presenting a challenge for detection. While machine learning shows promise, prior studies have not leveraged cardiovascular risk factors known impact hearing. As outcomes remain challenging characterize associations, we evaluated new approach predict current through using factors. National Health Nutrition Examination Survey (NHANES) 2012–2018 data comprising audiometric tests was utilized. Machine algorithms were trained classify impairment thresholds pure tone average values. Key results showed light gradient boosted performing best classifying mild or greater (> 25 dB HL) with 80.1% accuracy. It also classified > 16 HL 40 thresholds, accuracies exceeding 77% 86% respectively. study found that CatBoost Gradient Boosting performed well test set around 0.79 F1-scores 0.79–0.80. A multi-layer neural network emerged as the top predictor averages, achieving mean absolute error just 3.05 dB. Feature analysis identified age, gender, blood pressure waist circumference key associated Findings offer promising direction clinically applicable tool, personalized prevention strategies, calls prospective validation.

Language: Английский

Citations

0