Spoken Medical Prescription Acquisition Through a Dialogue System on Smartphone: Perspective of a Healthcare Software Company - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2020

Spoken Medical Prescription Acquisition Through a Dialogue System on Smartphone: Perspective of a Healthcare Software Company

Résumé

Industrial Medical Practice Management Software (PMS) have appeared in health institutions to reduce medication errors which affect several million people worldwide each year. However, practitioners must enter information manually into PMS which decreases the time devoted to care. In this paper, we describe the approach and some experiments of the implementation of an initial spoken dialogue system in this low-resourced domain in an industry-oriented setting. The main objective is to provide a natural language interface as an alternative to typing prescriptions in PMS. We highlight some of the difficulties of using deep-learning systems in an industrial context and discuss how these systems could be used while enabling full traceability. To overcome the lack of annotated speech data, we present a way to generate aligned data for machine-learning systems and discuss the limitations of using artificial data generation. We report on the findings of human evaluation conducted on the initial prototype with 2 medical experts and 2 naive users and discuss the results of each module of the dialogue system from an industrial perspective.
Fichier principal
Vignette du fichier
IndustryTrack_Reviewed_Kocabiyikoglu.pdf (168.36 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02996728 , version 1 (09-11-2020)

Identifiants

  • HAL Id : hal-02996728 , version 1

Citer

Ali Can Kocabiyikoglu, François Portet, Jean-Marc Babouchkine, Hervé Blanchon. Spoken Medical Prescription Acquisition Through a Dialogue System on Smartphone: Perspective of a Healthcare Software Company. LREC 2020 Industry Track Language Resources and Evaluation Conference 11–16 May 2020, Nov 2020, Marseille, France. ⟨hal-02996728⟩
107 Consultations
168 Téléchargements

Partager

Gmail Facebook X LinkedIn More