Support to Umeå University to develop AI tools for diagnosing acromegaly based on data from speech.
| Website: | AcroSpeech |
| PI: | Per Dahlqvist |
| Ongoing: | No |
| Start: | 2024-01-30 |
| End: | 2024-05-16 |
The AcroSpeech project aims to enable early diagnosis of acromegaly with the help of novel AI based voice recognition tools.
Acromegaly is a rare disease, caused by a benign tumor in the pituitary gland increasing productionen of growth hormone. This causes facial changes but also voice changes and swelling of soft tissues. The symptoms are unspecific and gradual, so therefore many can go on for as long as 10 years before getting a diagnosis. By that time, many have already developed bodily changes and other sequelae which may be permanent. Early diagnosis - and treatment - may therefore be life changing for the patients.
The AIDA Data Hub support function has hitherto been targeted mainly to the AIDA community in medical imaging diagnostics AI, but is now opening up provide user fee based support to a greater audience within precision health, and this is the both first support project outside of medical imaging and the first support project paid for by user fees. Here AIDA Data Hub helped with creating tools for human expert to annotate speech data which is essential to establish a baseline to compare AI performance agains. The tool was developed so that data was kept encrypted at all times, only being decrypted on the fly when annotators were presented with the data point.
If you are interested in receiving support for your project you can apply below.