An AI-Driven IoT Wearable Framework for Real-Time Health Monitoring and Early Risk Prediction in Lymphoma Patients Using Hybrid CNN-LSTM

Q: What is the clinical evidence from this study?

Population: Lymphoma (CLL, FL, MCL). Intervention: Hybrid CNN-LSTM and ResNet-50 with SAM framework vs. Baseline deep learning models (e.g., standard CNN, ResNet-50, VGG-16, MobileNetV2). Primary outcome: Classification accuracy for physiological time-series and histopathological images.

Q: What does this research mean for the field?

A hybrid CNN-LSTM and ResNet-50 framework integrated with IoT wearable sensors enables accurate real-time physiological risk prediction (96.1% accuracy) and histopathological classification (97.5% accuracy) for lymphoma patients. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Sebastián Santiago; Stalin Prince Chelladurai; G. Preethi; Subramanian Annamalai; Naramula Venkatesh; T. P. S. Kumar Kusumanchi

doi:10.22037/ijmtfm.v16.53262

September 25, 2026International Journal of Medical Toxicology and Forensic MedicineOpen Access

An AI-Driven IoT Wearable Framework for Real-Time Health Monitoring and Early Risk Prediction in Lymphoma Patients Using Hybrid CNN-LSTM

View Full Paper

Ask AI

Bookmark

Share

Key result

Hybrid AI framework achieves ~96% accuracy for physiological time-series analysis over standard deep learning models.

Why the study?

Lymphoma requires continuous monitoring of early symptoms of disease progression to enable prompt treatment, motivating development of real-time health monitoring frameworks.

Does a hybrid deep learning framework (CNN-LSTM and ResNet-50+SAM) improve the accuracy and speed of physiological monitoring and histopathological classification in lymphoma patients?

Key Points

Develop an AI-driven IoT wearable framework combined with deep learning architectures to provide continuous real-time monitoring and early risk prediction in lymphoma patients.
Designed an IoT framework capturing continuous physiological data—including heart rate, oxygen saturation (SpO₂), temperature, and electrocardiogram (ECG) metrics—coupled with a weighted risk scoring alert system.
Implemented a hybrid CNN-LSTM network for physiological time-series analysis and a ResNet-50 model with a Spatial Attention Mechanism (SAM) for histopathological classification across chronic lymphocytic leukemia, follicular lymphoma, and mantle cell lymphoma.
The hybrid CNN-LSTM model achieved an accuracy of 96.1%, precision of 94.7%, recall of 95.5%, F1-score of 95.1%, and an inference time of 7 ms for physiological time-series processing.
The ResNet-50 plus Spatial Attention Mechanism architecture reached a five-fold cross-validation mean accuracy of 97.50 ± 0.14% with an inference time of 6 ms for histopathological classification.

Structured PICO

Does a hybrid deep learning framework (CNN-LSTM and ResNet-50+SAM) improve the accuracy and speed of physiological monitoring and histopathological classification in lymphoma patients?

P

Population

A dataset of lymphoma histopathology images and physiological signals used to evaluate a deep learning-based IoT continuous monitoring framework.

I

Intervention

AI-driven IoT wearable framework using hybrid CNN-LSTM for physiological time-series analysis and ResNet-50 with Spatial Attention Mechanism (SAM) for histopathological classification

C

Comparator

Traditional deep learning models including CNN, ResNet-50, VGG-16, and MobileNetV2

O

Outcome

Model performance metrics including accuracy, precision, recall, F1-score, and inference timesurrogate

Main Result

Absolute Event Rate: 96.1% vs 87.5%

The AI-driven IoT framework demonstrates high accuracy and low inference latency for real-time physiological monitoring and histopathological classification in lymphoma patients.

Limitations

Further validation using larger, clinically annotated datasets is required before clinical deployment.
further validation using larger, clinically annotated datasets is required before clinical deployment

Implication

May enable real-time lymphoma monitoring; leaves open prospective validation before clinical adoption.

Abstract

Background: Lymphoma is a complex form of hematological cancer, and its early symptoms of disease progression need continuous monitoring for prompt treatment. The report proposes using the Internet of Things (IoT) for real-time patient monitoring, featuring wearable sensors, Deep Learning (DL) models, and an intelligent alert system to enhance predictive assessment of patient health. Methods: These smart wearable devices continuously measure physiological parameters, including heart rate, oxygen saturation (SpO₂), temperature, and electrocardiogram (ECG) parameters. A hybrid CNN-LSTM model is employed for physiological time-series analysis, while a ResNet-50 model enhanced with a Spatial Attention Mechanism (SAM) is used for histopathological classification of CLL, FL, and MCL. The reported performance of these models is evaluated separately according to their respective tasks. In addition, a weighted physiological risk score is used to identify abnormal physiological states from continuously monitored HR, SpO₂, temperature, and ECG measurements. An alert is generated when the risk score exceeds the predefined threshold. The framework is intended as an AI-assisted monitoring and decision-support system rather than an autonomous clinical diagnostic system. Results: The experimental results demonstrated that the proposed CNN-LSTM model achieved an accuracy of 96.1%, a precision of 94.7%, a recall of 95.5%, an F1-score of 95.1%, and an inference time of 7ms for the physiological time-series analysis. Additionally, ResNet-50+SAM had a remarkably low inference time of 6 ms and stable five-fold cross-validation results (mean accuracy: 97.50 ± 0.14%). Conclusion: This proposed framework offers numerous benefits to patients by enabling early detection of important health events and facilitating communication with healthcare professionals. This research demonstrates that artificial intelligence and IoT can transform real-time health monitoring, providing physicians with valuable information for personalized, ongoing care management of lymphoma patients.

Discussion

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

Cite This Study

Santiago et al. (2026) studied Lymphoma (CLL, FL, MCL). Hybrid CNN-LSTM and ResNet-50 with SAM framework vs. Baseline deep learning models (e.g., standard CNN, ResNet-50, VGG-16, MobileNetV2) was evaluated on Classification accuracy for physiological time-series and histopathological images. The proposed hybrid AI framework achieved a classification accuracy of 96.1% for physiological time-series analysis and 97.5% for lymphoma histopathological image classification.

synapsesocial.com/papers/6ac6e3c182aa59ee98b3fe91 https://doi.org/10.22037/ijmtfm.v16.53262

Topics

Artificial intelligence in cardiology

View Full Paper

Ask AI

Bookmark

Share

An AI-Driven IoT Wearable Framework for Real-Time Health Monitoring and Early Risk Prediction in Lymphoma Patients Using Hybrid CNN-LSTM

Key Points

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Topics

An AI-Driven IoT Wearable Framework for Real-Time Health Monitoring and Early Risk Prediction in Lymphoma Patients Using Hybrid CNN-LSTM

Key Points

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Keep reading

Topics

Keep reading