Homo Medicinalis. Recognising the voices of patients to retrace the opinions about medicines (HoMed)

Homo Medicinalis. Recognising the voices of patients to retrace the opinions about medicines (HoMed)

Applicant: dr. Henk van den Heuvel (Radboud University)

HoMed will implement a SSH research infrastructure with an enormous potential for automatic transcription of sensitive audio-visual (AV) recordings. Its use case will focus on AV-recordings of medical consultations on the use of pharmaceuticals (henceforth ‘MedPharm’). MedPharm in practice shows situations where patients often appear not to be able to understand proper medicine use. To overcome unintentional medicine use we need to better understand the attribution of meaning to medicines. Therefore, recordings and transcriptions of patient consultations are needed.

In HoMed a standard automatic speech recogniser (ASR) for Dutch will be adapted to MedPharm discourses, since essential jargon is not part of the vocabulary of the current generic ASR. The ASR will be retrained on existing radio and tv data and on highly sensitive AV-recordings of patient consultations at Nivel. The resulting infrastructure will be made available (i) as component of the ASR-service in CLARIAH’s Infrastructure (accessible via the Media Suite) and (ii) via Stichting OpenSpraaktechnologie to be used as a service tool in similar projects. In addition, a standalone version of the infrastructure developed for Nivel can be employed for sensitive data analysis under intramural conditions. The application will fully comply with the GDPR.

More information