Synthetic Photoplethysmogram (PPG) Generation Using Genetic Programming Based Generative Model
Subject Areas : هوش مصنوعی و رباتیکFatemeh Ghasemi 1 , Fardin abdali 2
1 - Department of Computer Engineering and Information Technology, Razi University, Kermanshah, Iran
2 -
Keywords: Photoplethysmogram, Generative Model, Genetic Programming, Scalability, Mathematical Model.,
Abstract :
Today, advancements in Information and Communication Technology (ICT), particularly in healthcare and cardiac activity monitoring, have led to increased adoption of Photoplethysmogram (PPG) technology in smart devices and mobile phones. The development of generative models for producing artificial PPG signals faces challenges such as a lack of diversity and constraints in training data. This article employs a Genetic Programming (GP) based approach to introduce a generative model capable of producing PPG data with increased diversity and enhanced accuracy using an initial sample of PPG signals. In contrast to conventional regression, Genetic Programming automates the determination of the mathematical model's structure and compositions. The proposed approach, with a Mean Squared Error (MSE) of 0.0001, Root Mean Square Error (RMSE) of 0.01, and a correlation of 0.999, demonstrates superior performance due to appropriate optimization and acceptable accuracy in generating synthetic PPG data. It outperforms other methods in terms of efficiency and execution capability, especially in resource-constrained environments.