Skip to content

Medical Devices Digital Twins with Meta-Learning

Notifications You must be signed in to change notification settings

Simula-COMPLEX/MeDeT

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

15 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

MeDeT: Medical Devices Digital Twins Generation with Meta-learning

The MeDeT approach focuses on building, adapting, and operating high-fidelity digital twins (DTs) of medical devices, employing few-shot meta-learning techniques. These medical devices DTs are designed to streamline testing automation for healthcare IoT applications.

MeDeT works in six phases: (i) Data Generation - generates raw data for medical devices, (ii) Data Preparation - preprocesses raw data for training, (iii) Meta-learning - creates meta dataset & taskset, determines model architecture, and trains/fine-tunes with MAML algorithm, (iv) Build DTs - creates model clones, storage, APIs, and JSON objects, (v) DT Request Handler - processes requests from a healthcare IoT application during testing, and (vi) DTs to Device Communication - establishes DTs communication with medical devices.

This work is a part of the Welfare Technology Solution (WTT4Oslo) project (#309175) funded by the Research Council of Norway.

Basic Requirements

  • Machine: minimum 16GB RAM and 8-core processor
  • OS: MacOS or Windows 10
  • IDE: PyCharm
  • Python: 3.8 or higher

Dependencies

  • PyTorch: 2.0.1
  • learn2learn: 0.2.0
  • scikit-learn: 1.3.0
  • Pandas: 2.0.3
  • Flask: 2.2.3
  • Flask-RESTful: 0.3.9

Releases

No releases published

Packages

No packages published

Languages