AI-Enabled Digital Twin for Smart Healthcare Monitoring

Q: What is the clinical evidence from this study?

Population: Cardiac arrhythmia (n=87554). Intervention: Edge-Intelligent AIoT Digital Twin Framework with Random Forest Classifier vs. Continuous transmission-based cloud monitoring model. Primary outcome: Overall ECG heartbeat classification accuracy.

Q: What does this research mean for the field?

An edge-intelligent, event-driven digital twin framework reduces transmitted ECG data volume by over 99% and device energy consumption by approximately 85.6% while achieving a 98.47% heartbeat classification accuracy. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

doi:10.58257/ijprems55049

October 22, 2026International Journal of Progressive Research in Engineering Management and ScienceOpen Access

AI-Enabled Digital Twin for Smart Healthcare Monitoring

View Full Paper

Ask AI

Bookmark

Share

Key result

Edge-intelligent AIoT digital twin achieves ~98% ECG classification accuracy while cutting data transmission over 99%.

n=87,554

Why the study?

Cloud-based healthcare monitoring systems continuously transmit physiological sensor data to remote servers, causing large communication overhead, energy consumption, processing delays, and potential patient privacy violations.

Does an Edge-Intelligent AIoT Digital Twin framework using a Random Forest classifier maintain accurate ECG arrhythmia detection while reducing communication overhead and energy consumption compared to continuous cloud transmission?

Key Points

To develop an edge-intelligent digital twin framework that detects cardiac events locally and transmits event-driven updates rather than continuous raw sensor data.
Designed a four-layer architecture incorporating sensing, edge intelligence, digital twin management, and cloud services using a Random Forest Classifier to identify cardiac events.
Tested event detection accuracy on the MIT-BIH Arrhythmia Dataset containing 87,554 labeled heartbeat samples across five categories.
Evaluated communication efficiency, energy consumption, and scalability for monitoring up to 100 patients against a continuous raw ECG transmission baseline.
The Random Forest Classifier achieved an overall accuracy of 98.47%, a normal heartbeat sensitivity of 99.56%, and a ventricular heartbeat sensitivity of 93.45%.
Event-driven messaging reduced transmitted data volume from 62.208 MB/day to 289 KB/day per patient, representing an estimated reduction of approximately 99.54% (and up to 99.95% against expanded baselines).
Daily energy consumption dropped from 420 mWh to 60 mWh (an 85.6% to 85.7% reduction), increasing estimated wearable battery life from 4.4 days to 30.7 days.

Structured PICO

Does an Edge-Intelligent AIoT Digital Twin framework using a Random Forest classifier maintain accurate ECG arrhythmia detection while reducing communication overhead and energy consumption compared to continuous cloud transmission?

P

Population

87,554 labeled ECG heartbeat samples from the MIT-BIH Arrhythmia Database categorized into five heartbeat classes (88.5% normal beats).

I

Intervention

Edge-Intelligent AIoT Digital Twin Framework incorporating a four-layer architecture (sensing, edge intelligence, digital twin management, cloud services), a Random Forest Classifier using 20-24 time- and frequency-domain engineered ECG features, and event-driven semantic message transmission.

C

Comparator

Continuous-transmission baseline cloud monitoring model relaying raw 360 Hz ECG measurements (62.208 MB/day or 622 MB/day baseline) to a central server.

O

Outcome

Overall ECG heartbeat classification accuracy and class-specific sensitivity and F1-score.surrogate

Combining edge-based Random Forest ECG classification with event-driven Digital Twin synchronization achieves 98.47% arrhythmia classification accuracy while drastically reducing data transmission and extending wearable battery life.

Limitations

Large class imbalance in the dataset with nearly 88.5% of heartbeats classified as normal
Insufficient data for the unclassifiable beat category (only 8 instances), resulting in 50.00% sensitivity
Requires patient-independent data splitting into training and testing datasets to assess generalization across different patients
Energy and communication analyses are model-based estimates requiring physical hardware testing for validation
Confidence and persistence thresholds for event detection still need clinical evaluation for targeted patient populations
Dataset is heavily imbalanced with nearly 88.5% normal heartbeats
Very rare unclassifiable beat class (only 8 instances) yielded uninformative 50.00% sensitivity and F1-score
Lack of patient-independent data splitting in current evaluation to prove generalizability across different patients
Energy consumption and communication reductions are model-based calculations without physical hardware testing
Clinical criteria, confidence thresholds, and persistence thresholds require clinical validation in target patient populations

Ask about this paper

Answers are grounded in this paper and cite their sources.

Implication

May support efficient wearable ECG monitoring; leaves open prospective validation for clinical arrhythmia outcomes.

Abstract

The growing number of people suffering from chronic diseases along with the requirement for constant monitoring of patients' health status and the development of AI, IoMT, Edge Computing and Digital Twin (DT) technologies have contributed to rapid implementation of these technologies.On the other hand, Cloud-Based Healthcare Monitoring Systems continuously transmit physiological sensor data to remote servers, causing large communication overhead, energy consumption, processing delays, and potentially violating patients' privacy rights.To solve these problems, this study proposes an Edge-Intelligent AIoT Digital Twin Framework that incorporates machine learning-based health event detection, event-driven communication, and dynamically updated virtual patient representation to minimize communication and maintain patient privacy.The proposed framework has a four-layer architecture including sensing, edge intelligence, digital twin management and cloud services.Random Forest Classifier examines Electrocardiogram (ECG) signals in order to identify local clinically relevant cardiac events.Instead of sending continuous streams of raw ECG signals, the framework transmits small semantic event messages that update the patient's Digital Twin model.The framework is tested using the MIT-BIH Arrhythmia Dataset containing 87,554 labeled heartbeat samples divided into five heartbeat categories.Thesis presents an overall classification accuracy of 98.47% and ventricular heartbeat sensitivity of 93.45%.The event-driven communication model reduces the amount of data transmitted by approximately 99.95 percent compared to a continuous transmission-based model.Model-Based Energy Analysis shows an approximate 85.6 percent decrease in daily energy use which increases the estimated battery life from 4.4 days to 30.7 days.Scalability analysis evaluated the performance of the model for monitoring up to 100 patients.Results show that combining Edge Intelligence and Event-Driven Digital Twin Synchronization will improve communication efficiency while maintaining critical information about patient health status.The proposed model provides a basis for developing scalable, resourceefficient, and privacy-aware remote healthcare monitoring applications.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

A 2026 study studied Cardiac arrhythmia (n=87,554). Edge-Intelligent AIoT Digital Twin Framework with Random Forest Classifier vs. Continuous transmission-based cloud monitoring model was evaluated on Overall ECG heartbeat classification accuracy. The Edge-Intelligent AIoT Digital Twin framework achieved 98.47% overall ECG classification accuracy while reducing daily data transmission by 99.54% to 99.95% and energy use by 85.6%.

synapsesocial.com/papers/6af10ed2a61b6e74a6de1584 https://doi.org/10.58257/ijprems55049

View Full Paper

Ask AI

Bookmark

Share

AI-Enabled Digital Twin for Smart Healthcare Monitoring

Key Points

Structured PICO

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

AI-Enabled Digital Twin for Smart Healthcare Monitoring

Key Points

Structured PICO

Limitations

Ask about this paper

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading