Exploring linguistic feature and model combination for speech recognition based automatic AD detection

TL;DR

This paper enhances automatic Alzheimer's detection from speech by combining linguistic features and models, using fine-tuned pre-trained encoders and ensemble classifiers, achieving state-of-the-art accuracy on a benchmark dataset.

Contribution

It introduces a novel combination of feature and model fine-tuning strategies for improved robustness in AD detection from limited speech data.

Findings

Achieved 91.67% accuracy with manual transcripts.

Achieved 93.75% accuracy with ASR transcripts.

Demonstrated robustness of model and feature combination methods.

Abstract

Early diagnosis of Alzheimer's disease (AD) is crucial in facilitating preventive care and delay progression. Speech based automatic AD screening systems provide a non-intrusive and more scalable alternative to other clinical screening techniques. Scarcity of such specialist data leads to uncertainty in both model selection and feature learning when developing such systems. To this end, this paper investigates the use of feature and model combination approaches to improve the robustness of domain fine-tuning of BERT and Roberta pre-trained text encoders on limited data, before the resulting embedding features being fed into an ensemble of backend classifiers to produce the final AD detection decision via majority voting. Experiments conducted on the ADReSS20 Challenge dataset suggest consistent performance improvements were obtained using model and feature combination in system development. State-of-the-art AD detection accuracies of 91.67 percent and 93.75 percent were obtained using manual and ASR speech transcripts respectively on the ADReSS20 test set consisting of 48 elderly speakers.