Acoustic Measures in Diagnosis and Subtyping of Childhood Apraxia of Speech

ABSTRACT Childhood apraxia of speech (CAS) is a pediatric speech sound disorder (SSD) that impacts between 60,000 and 780,000 children under 14 in the US. The underlying impairment in speech motor planning primarily affects articulation and prosody and reduces intelligibility. Beyond impaired communication, children diagnosed with CAS are also at higher risk for difficulties with social/emotional development and reading and writing. Thus, effective targeted intervention is warranted for these children. However, CAS diagnosis is difficult and at present relies primarily on the `gold standard' clinical judgment by an `expert speech-language pathologist (SLP)' based on a number of perceptual features: (a) Inconsistent errors, (b) impaired transitions between sounds and syllables, and (c) impaired prosody, in particular lexical stress. This diagnostic standard is problematic for several reasons, including the fact that these features are not operationalized (in terms of elicitation, measures, number of instances required for diagnosis), lack precision, and vary with the SLP's language and clinical experience. These challenges are exacerbated by the well-known heterogeneity of CAS presentation and the overlap with other SSDs. Finally, there are no standards or criteria by which an SLP is considered an `expert'. Together, these major limitations lead to disagreements between SLPs and reluctance by some SLPs, especially less experienced ones, to diagnose CAS. This undermines the ability to make well-informed evidence-based clinical decisions for children with CAS and other SSDs. In addition, the research literature on CAS has been hampered for decades by this diagnostic problem, and is compounded by small sample sizes. More objective, validated, precise, replicable, and clinically feasible diagnostic measures are sorely needed. Acoustic measures have strong potential to meet these needs, based on prior small-scale studies in CAS and studies with other populations with speech disorders. However, to date no studies have systematically examined acoustic measures of core perceptual CAS features in a single study. This F31 proposal leverages a unique, large and well-characterized retrospective dataset of 125 children (76 with CAS, 28 with another SSD, and 21 without any SSD) to systematically examine the validity of a well- motivated set of acoustic measures of each core perceptual feature (3 acoustic measures each) (Aim 1), the differential diagnostic accuracy of the three acoustic measures with highest validity (1 acoustic measure per feature) (Aim 2), and the potential identification of CAS subtypes based on these acoustic measures (Aim 3). As a first step, the diagnostic accuracy of these acoustic measures will be evaluated against a gold standard of an operationalized consensus criterion based on independent judgments from 3 expert SLPs. This F31 project will provide novel, unique, and significant information that will help improve clinical diagnosis of CAS and other SSDs, provide the Fellow with advanced training to transition toward an independent research career, and set the stage for a future prospective validation study of the acoustic measures found most promising here.