Voice Analysis in Asthmatic Patients With Machine Learning Models

Voice Analysis in Asthmatic Patients and Healthy Individuals: Comparative Evaluation of Asthma Control Levels and Voice Characteristics With Machine Learning Models

Asthma can lead to various factors that impair voice production, including airway restriction, inflammation, and mucus production, resulting in changes in voice frequency and amplitude. Therefore, voice analysis may serve as an indicator of respiratory diseases.

A national, observational, case-control study is planned in Türkiye to analyze differences in voice between healthy subjects and asthmatic patients and to assess voice analysis techniques for determining an effective biomarker for asthma control using a machine learning model.

Asthma is a disease characterized by chronic inflammation. Based on the frequency of symptoms and the use of reliever medications, the disease can be classified as either 'controlled' or 'uncontrolled'. Currently, GINA criteria and Asthma Control Test can be used to evaluate asthma control.

The relationship between respiratory functions and speech has been previously studied, revealing that voice changes can occur in asthmatic patients due to symptom presence. Asthma can lead to various factors that impair voice production, including airway restriction, inflammation, and mucus production, resulting in changes in voice frequency and amplitude. Therefore, voice analysis may serve as an indicator of respiratory diseases. Understanding the alterations in phonation/voice due to the underlying disease is crucial.

This study seeks to analyze differences in voice between healthy subjects and asthmatic patients and to assess voice analysis techniques for determining an effective biomarker for asthma control using a machine learning model.

This is a national, observational, cross-sectional study that will be conducted in Türkiye. The study consists of two stages: in the first stage, a machine learning (ML) model will be developed using voice data collected from both healthy individuals and patients diagnosed with asthma. In the second stage, this ML model will be tested to detect voice differences among patients at different levels of asthma control.

Asthmatic Group

Inclusion Criteria:

• Patients diagnosed with asthma according to GINA criteria and Pulmonary Function Test, and followed for at least three months
• 18-65 years of age.
• Sign an informed consent document
• Able to comply with the study protocol during the study period.

Exclusion Criteria:

• None

Healthy Group

Inclusion Criteria:

• Healthy participants between 18-65 years of age
• Good general health
• No history of chronic respiratory disorders
• No history of chronic systemic disorders
• No history of upper respiratory tract infections within five days prior voice recording.

Exclusion Criteria:

• None