ABSTRACT
This article describes the design of KommPaS, a web-based tool for the clinical assessment of communication impairment in persons with dysarthria. KommPaS (the German acronym for Communication-related Parameters in Speech Disorders) allows clinicians to recruit laypersons via crowdsourcing for the evaluation of samples of dysarthric speech with regard to communication relevant parameters, that is, intelligibility, naturalness, perceived listener effort, and efficiency (intelligible speech units per unit time). Moreover, a communication total score describing the KommPaS profile elevation, i.e., the arithmetic mean of the normalized KommPaS scores, is provided. Based on considerations regarding the theoretical underpinnings and methodological constraints of a clinical tool for the assessment of these parameters, the article describes how each theoretically and methodologically motivated feature is translated into design principles and how these principles are implemented in a web application. The paper reports efficiency data and details the data privacy and data security provisions that are essential in such an approach.
Acknowledgments
This work was funded by the German Academic Scholarship Foundation and the Bayerische Sparkassenstiftung. Preliminary work was supported by an ERC proof-of-concept grant awarded to Jonathan Harrington (ERC-POC 737552). We are deeply grateful to Klaus Jänsch, Christoph Draxler and Raphael Winkelmann for technical support.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 The database is freely available as an excel document (https://neurophonetik.de/subtlex-np).
2 Ethical approval (Project No. 19–365) has been obtained from the Ethics Committee of the Faculty of Medicine at the Ludwig-Maximilians-University Munich, Germany. The patients were informed in detail about the study by the examiner and gave their written consent to participate.
3 For anonymization all personal data, including the speech samples, are deleted and only the numerical results and a non-personal, unpredictable ID are stored.