An Introduction to Machine Learning for Speech-Language Pathologists: Concepts, Terminology, and Emerging Applications.

Abstract

Purpose: The purpose of this article is to orient both clinicians and researchers to machine learning (ML) approaches as applied to the field of speech-language pathology. We first introduce key ML concepts and terminology and proceed to feature exemplar papers of recent work utilizing ML techniques in speech-language pathology. We also discuss the limitations, cautions, and challenges to the implementation of ML and related techniques in speech-language pathology.

Conclusions: Readers are introduced to broad ML concepts, including common ML tasks (e.g., classification, regression), and specific types of ML models (e.g., linear/logistic regression, random forest, support vector machines, neural networks). Key considerations for developing, evaluating, validating, and interpreting ML models are discussed. An application section reviews six exemplar published papers in the aphasiology literature that have utilized ML approaches. Lastly, limitations to the implementation of ML approaches are discussed, including issues of reliability, validity, bias, and explainability. We highlight emergent solutions and next steps to facilitate responsible and clinically meaningful use of ML approaches in speech-language pathology moving forward.