Datasets
Standard Dataset
Fully automated insulin treatment
- Citation Author(s):
-
SeunghyunLee
POSTECH - Submitted by:
- Seunghyun Lee
- Last updated:
- Fri, 05/01/2020 - 06:10
- DOI:
- 10.21227/z0nt-z923
- Data Format:
- License:
1188 Views- Categories:
- Keywords:
Abstract
This dataset contains in-silico results of insulin treatment using a fully automated artificial pancreas algorithm based on reinforcement learning for FDA-approved virtual patients (C. D. Man et al., 2014) with type 1 diabetes (10 adults and 10 adolescents).
The data contains the treatment results of the trained reinforcement learning based control algorithm over three different random seeds in each of the 20 virtual patients on seven runs of a single-meal (SM) scenario (randomly gernerated from mean amount of 65g with a standard deviation of 17g) following a preprandial fasting period with measurement noise. It also contains the treatment results of the trained algorihtm over three runs on two days of real life micking mult-meal (MM) scenario (40 g of CHO for breakfast at 8:00 a.m., 80 g of CHO for lunch at 1:00 p.m., and 60g of CHO for dinner at 9:00 p.m.).
Comments
...