Predicting glucose from peripheral nerve signals
Dnr:
NAISS 2023/22-1287
Type:
NAISS Small Compute
Principal Investigator:
Oskar Allerbo
Affiliation:
Kungliga Tekniska högskolan
Start Date:
2023-12-13
End Date:
2025-01-01
Primary Classification:
10106: Probability Theory and Statistics
Webpage:
Allocation
- Mimer at C3SE: 500 GiB
- Klemming at PDC: 500 GiB
- Alvis at C3SE: 250 GPU-h/month
- Dardel-GPU at PDC: 250 GPU-h/month
- Dardel at PDC: 5 x 1000 core-h/month
Abstract
I hold a postdoc position in mathematical statistics at KTH, financed by WASP-DDLS (Wallenberg Autonomous Systems and Software program / Data-Driven Life Science), where the overall aim of the project is to decode nerve signals for prediction of glucose in the blood. We have had some initial success with deep models for sequential data, more specifically by training LSTMs (long short-term memory models) on a laptop, and believe that additional computational resources will help to facilitate the research.