An Automatic Assessment System for Alzheimer's Disease Based on Speech Using Feature Sequence Generator and Recurrent...

Area of Science:

• Neurology
• Artificial Intelligence
• Speech Science

Background:

• Dementia, including Alzheimer disease, is a leading global cause of death.
• Early detection is vital for timely intervention due to the lack of a cure.
• Speech analysis offers a cost-effective and accessible method for cognitive assessment compared to medical imaging or blood tests.

Purpose of the Study:

• To develop an automated speech analysis system for early dementia detection.
• To propose a novel Feature Sequence representation for speech data.
• To utilize a data-driven approach for classification.

Main Methods:

• A novel Feature Sequence representation was developed for speech data.
• Recurrent neural networks (RNNs) were employed for automated classification.
• The system was designed for full automation and wide deployment.

Main Results:

• Experiments were conducted using 120 speech samples.
• The automated system achieved a high performance score.
• The area under the receiver operating characteristic curve (AUC) reached 0.838.

Conclusions:

• Automated speech analysis using RNNs and Feature Sequence representation is a promising tool for dementia detection.
• The system's automated nature facilitates widespread accessibility for public health.
• The high AUC score validates the effectiveness of this approach for cognitive assessment.