How Mabel Works
The app is intended to be available on the health care provider’s mobile devices. All AI and functionality stays in the phone and no internet connection is needed. Nothing leaves the phone, or the room.
The health care provider starts the session in the app by just talking. The app captures the voice and converts it into text using machine learning.
The text is sent to machine translation which generates the translation to your target language instantly. The text can now be played as speech in the target language.
The app take the inverse path when it’s the patient’s turn and seamlessly supports the conversation cross language barriers.
The whole patient/HCP session is available as a transcript and can be sent directly to the Electronic Patient Journal system over a secure connection. No trace of the conversation will be left on the phone, securing full confidentiality.
We train and validate our speech models by taking medical terminologies and sentences and introducing a lot of background noises, such as crowds, hospital corridors, people speaking a different language, and even artillery fire (the solution is intended to be used in Ukraine).