MAIA.Tools: Medical artificial intelligence assistance tools

The project supports automatic analysis of conversations between doctors and patients such as explanations about treatments, surgeries, or aftercare.

The MAIA-ASR system records the conversation, analyzes and assesses it contextually and semantically, and transfers it into written documentation. We are therefore not only actively involved in the development of the basic technology, but also focused on a transition into a marketable product.

Every speech recognition application is based on a well-functioning speech-to-text technology. In the medical domain, in spontaneous doctor-patient conversations, especially in German, the speech-to-text accuracy of the market-leading ASR providers is not satisfactory. Furthermore, the available analysis routines for such texts are not made for the special cases of documenting patient conversations in semantic and lexical terms.

Together with the MAIA.TOOLS, we are working thus on an improvement of both the ASR technology and the text analysis. As in every development in the field of ASR, we need recordings of real conversations. We record these with completely new microphones from our hardware partners Philips SPS. In the automatic analysis of doctor-patient conversations, we apply the state-of-the-art NLP and IE methods, a set of dedicated, medical domain-specific linguistic resources, and medical procedure ontologies. On the medical side, we rely on partners in the private practice sector as well as partners in clinics to provide us with the necessary data.

ASR - Automatic Speech Recognition
NLP - Natural Language Processing
IE - Information Extraction

Researchers: Brigitte Krenn, Marcin Skowron
Contact (OFAI): brigitte.krenn@ofai.at
Contact (MAIA.TOOLS, Johannes Allesch): johannes@maia.tools

project website