Accelerating emerging AI language models
The software works with speech recognition and AI language models. And although the development of the software had been in progress for a number of years, it accelerated at the beginning of last year. This had to do with the arrival of GPTs, including ChatGPT. “When we saw what these models can do, we thought: this is exactly what we need,” says Brinkkemper. “It is the semantic technology needed to convert the transcriptions that speech recognition produces into the information that the EPD requests.” “But we also immediately saw the dangers of this large languagemodels, such as misinterpreting words or sentences,” adds Thymen den Hartog. He is one of the former students and co-founders of Verticai. “The models really need to be framed to understand the context of different specialties.”
Collaboration partners in healthcare
To achieve this, Verticai has now run a number of pilots with partners in healthcare. Together they train the software with sample consultations so that it generates increasingly better reports. Moreover, the information requirements are not the same for the different specialties. For example, the information that a cardiologist must provide for the Electronic Patient File is very different from the information that a neurologist or orthopedist is asked for. Based on the guideline of the medical specialty, this is also being worked on.
Collaboration partners include the Geriatrics department of the Radboud UMC, a number of physiotherapy practices in the Arnhem and Nijmegen area and an institution specialized in the treatment of PTSD. “We get a lot of pleasure from further expanding our innovative system together with healthcare workers,” says Den Hartog enthusiastically, “so that they can ultimately fully focus on what they enjoy doing most: providing dedicated care.”
If you would like more information here, please contact Verticai at [email protected]
Tags: Spinoff Utrecht University eases administrative burden healthcare professionals News
