The Frankestein project

Simulink project to provide pressure and flow values in the human arteries assisted by AI.

Brief description

This is a Simulink project to model the human arteries using electric circuit components and assisted by an AI machine. The Simulink model uses the reported circuit diagram for the human arteries in [1], and for the heart the electric circuit model in [2, Fig. 4.2.1 pag. 79]. The AI implements a DQN model, that try to adjust generated pressure values to a given database sequence. Further details on this design are given in [3]

How to run this project

The project runs directly from the live script file 1_training_and_deployment_noordergraaf_rideout_AI.mlx. This file runs the code to load the circuit elements and train the DQN Agent while running the simulink design. Although the simulink file can be oppened the project only runs from the above mlx file.

Project files

• Live script file "A_training_and_deployment_noordergraaf_rideout_AI.mlx": This is the master file to run the project. This file consist of four main blocks. The first block is to load the parameters and files. The second block is for the DQN Agent training. The third block performs the deployment, and the last block plot results.
• Mat file "out_noordergraaf_rideout.mat": This file stores the ground truth for the ML training.
• Live script file "Electric_Circuit_Parameters.mlx": This files load the electric circuit parameters of the Simulink design. On the one hand, it loads all the parameters from [2] are defined in the file init_fcn_noordergraaf_rideout.mlx, which are given in the CGS units. On the second hand, it reads the paramaters for the arteries in the excel table "Parameters.xlsx", these parameters are specified with the International System of Units (SI) accoding to [1].
• Excel table "Parameters.xlsx": This file includes all the electric parameters for the arteries following the original design in [1].
• Simulink file "Noordergraaf_Rideout_AI_Aorta_Ascendens.slx": Comprises the design for the electric circuit model and its integration to the ML module.

References

[1] Abraham Noordergraaf, Pieter D. Verdouw, and Herman B.K. Boom. 1963. The use of an analog computer in a circulation model. Progress in Cardiovascular Diseases 5, 5 (3 1963), 419–439. https://doi.org/10.1016/s0033-0620(63)80009-2

[2] Vincent C. Rideout. 1991. Mathematical and Computer Modeling of Physiological Systems. Prentice Hall, Upper Saddle River, NJ.

[3] Jorge Torres Gomez, Nicolai Spicher, Jorge Luis González Rios, and Falko Dressler. 2023. Fine-tuned Circuit Representation of Human Vessels through Reinforcement Learning: A Novel Digital Twin Approach for Hemodynamics. In Proceedings of the 10th ACM International Conference on Nanoscale Computing and Communication (NANOCOM '23). Association for Computing Machinery, New York, NY, USA, 46–52. https://doi.org/10.1145/3576781.3608717