Robust Non-Invasive Cardiac Index Prediction via Feature Integration and Data-Augmented Neural Networks

Chih-Hao Chang¹, Mei-Ling Chan², Yu-Hung Fang³

¹Department of Computer Science and Information Engineering, National Central University, Taoyuan 320317, Taiwan.

|May 4, 2026

Summary

This study uses non-invasive Internet of Things (IoT) devices and an artificial neural network (ANN) to predict cardiac index (CI). The ANN model achieved 97.78% accuracy, offering a practical solution for real-time cardiovascular health assessment.

Area of Science:

Cardiovascular Health
Biomedical Engineering
Artificial Intelligence in Medicine

Background:

Rising obesity and cardiovascular issues in young adults due to lifestyle changes present a global health challenge.
Current methods for assessing cardiac hemodynamic parameters can be invasive or lack real-time capabilities.
There is a need for accessible, non-invasive methods for monitoring cardiovascular health in young adults.

Purpose of the Study:

To develop and validate a non-invasive framework for predicting cardiac index (CI) using Internet of Things (IoT) sensing devices and artificial neural networks (ANN).
To assess the accuracy and stability of the proposed ANN model for real-time CI prediction.
To explore the potential of this integrated system for early detection and management of cardiovascular conditions.

Main Methods:

Integration of non-invasive IoT devices: TERUMO ES-P2000 blood pressure monitor and PhysioFlow PF07 Enduro cardiac hemodynamic analyzer.
Development of an artificial neural network (ANN) model for cardiac index (CI) prediction, utilizing physiological parameters.
Application of data preprocessing and advanced model training strategies to enhance prediction accuracy and model generalization.

Main Results:

The ANN model achieved a high classification accuracy of 97.78% when using three physiological parameters as input.
The model demonstrated strong predictive performance even with only two input parameters.
The proposed framework proved effective and practical for real-time, non-invasive CI assessment, outperforming traditional methods.

Conclusions:

The developed ANN-based framework offers a highly accurate and non-invasive method for real-time cardiac index assessment.
This technology holds significant potential for improving cardiovascular health monitoring and management in young adults.
The study highlights the successful application of IoT and AI in addressing critical public health concerns related to cardiovascular disease.

Keywords:

IoMT IoT artificial neural networks cardiac index cardiovascular disease assessment data augmentation machine learning non-invasive sensing physiological signal processing

Related Experiment Videos

Robust Non-Invasive Cardiac Index Prediction via Feature Integration and Data-Augmented Neural Networks

Related Concept Videos

Benign Share Benefit to Malignant: Balanced Mixing on Feature Space for Imbalanced Breast Cancer Classification.

Dual temperature- and magnetic-responsive PNIPAAm-based nanogels for tunable and drug release.

Bispecific antibody AP402 targets p95HER2 and induces costimulation through CD137 to improve efficacy for HER2-positive tumors.

Quantitative measurement of agonistic behaviors of intra- and interspecifics of Gryllus bimaculatus and Acheta domesticus by using DeepLabCut.

Pediatric intensive care unit nurses' reflective experience in caring for maltreated children and adolescents in Taiwan: A narrative qualitative study.

Species-specific primers in multiplex PCR for rapid identification of the biting midge Forcipomyia taiwana (Diptera: Ceratopogonidae).

RETRACTED: Alam et al. Empirical Study of Autism Spectrum Disorder Diagnosis Using Facial Images by Improved Transfer Learning Approach. Bioengineering 2022, 9, 710.

Correction: Apa et al. Ultrasonic Surgical Aspirator in Intramedullary Spinal Cord Tumours Treatment: A Simulation Study of Vibration and Temperature Field. Bioengineering 2025, 12, 842.

Correction: Peng et al. CT-Based Habitat Radiomics Combining Multi-Instance Learning for Early Prediction of Post-Neoadjuvant Lymph Node Metastasis in Esophageal Squamous Cell Carcinoma. Bioengineering 2025, 12, 813.

Surface Electromyography-Based Motion Analysis of Thigh Muscle Activation During the Modified Star Excursion Balance Test in Novice Recreational Runners with Chronic Ankle Instability: A Preliminary Cross-Sectional Case-Control Study.

State-of-the-Art Review on Finite Element Analysis Applied to Ultrathin-Strut Drug-Eluting Stents: Examining the Present and Looking Ahead to the Future.

Mapping Neuroanatomical Heterogeneity of Brain Aging Within a Clinically Defined Aging Reference Cohort.

Free-space coherent optical dot-product multiplier with lensless fan-in.

Photonic transcendental equation solver using frequency to time mapping.

Adaptive pinhole point diffraction interferometer based on a liquid crystal on silicon spatial light modulator.

Photonic-electronic hybrid distributed acoustic sensing for broadband detection systems.

Diffractive optoelectronic-reservoir computing with intrinsic spatiotemporal dynamics.

Thin and soft optical tactile sensor for highly sensitive object perception.

Related Experiment Videos

Robust Non-Invasive Cardiac Index Prediction via Feature Integration and Data-Augmented Neural Networks

Area of Science:

Background:

Purpose of the Study:

Main Methods:

Main Results:

Conclusions:

Related Concept Videos

Related Articles