Precision Cardiology: Leveraging Cognitive Computing Within Human Digital Twin Framework

Cardiovascular disease persists to be one of the leading mortality factors across the globe. Our research for precise and personalized prognosis of human cardiovascular risk with the novel Human Digital Twin framework is a step towards alleviating the risk for humankind, which profoundly aligns with the 3rd goal of UN-SDG 2030. The proposed framework combines primary echo and demographic data collected from Touch Hospital, Telangana, India with advanced cognitive computation methods for efficient prediction of cardiac risk. The first step in our experimentation involves creating a flexible Recurrent Generative Adversarial Network (R-GAN) architecture for echo data augmentation in order to overcome the limited availability of primary data. The R-GAN with GRU in the discriminator and LSTM in the generator results in the production of synthetic data of the highest quality. In the next step, we are building a State-Space Model (SSM) to analyze the temporal cardiovascular risk behavior with a transition function powered by four classifiers: XGBoost, Vanilla-RNN, LSTM, and GRU. The SSM also ensures the incorporation of a feedback loop through temporal adaptability with the Kalman Filter. The SSM with LSTM transition function is able to produce the most effective cardiac risk prediction result with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 8. 1 3 \%}$</tex> accuracy. This SSM has also been evaluated and proved to be free from consequential overfitting bias.