OpenAlex · Aktualisierung stündlich · Letzte Aktualisierung: 23.05.2026, 10:51

Dies ist eine Übersichtsseite mit Metadaten zu dieser wissenschaftlichen Arbeit. Der vollständige Artikel ist beim Verlag verfügbar.

Care for Context: Evaluating Speech-to-Text and Translation Services for 24h Live-In Care Communication

2026·0 Zitationen·Studies in health technology and informaticsOpen Access
Volltext beim Verlag öffnen

0

Zitationen

3

Autoren

2026

Jahr

Abstract

BACKGROUND: The 24h live-in care sector relies on Eastern European workers and multilingual communication. OBJECTIVES: The aim was to select the best AI-based language services for a 24h care communication app under real deployment constraints. METHODS: A scenario-based quality comparison for transcription, translation, and their cascaded pipeline, and an analysis of the risks and harmfulness for the specific application context were performed. RESULTS: Chirp2 and DeepL showed the best quality. We balanced quality with latency and detected the presence of potentially harmful meaning shifts. CONCLUSION: Domain-grounded evaluation provides a solid base for informed engineering decisions, but considering constraints and risks is necessary for end-to-end safety in deployment.

Ähnliche Arbeiten

Autoren

Institutionen

Themen

Artificial Intelligence in Healthcare and EducationInterpreting and Communication in HealthcareElectronic Health Records Systems
Volltext beim Verlag öffnen