Deep Learning-Based IoT System for Remote Monitoring and Early Detection of Health Issues in Real-Time

Sensors, Journal Year: 2023, Volume and Issue: 23(11), P. 5204 - 5204

Published: May 30, 2023

With an aging population and increased chronic diseases, remote health monitoring has become critical to improving patient care reducing healthcare costs. The Internet of Things (IoT) recently drawn much interest as a potential remedy. IoT-based systems can gather analyze wide range physiological data, including blood oxygen levels, heart rates, body temperatures, ECG signals, then provide real-time feedback medical professionals so they may take appropriate action. This paper proposes system for early detection problems in home clinical settings. comprises three sensor types: MAX30100 measuring level rate; AD8232 module signal data; MLX90614 non-contact infrared temperature. collected data is transmitted server using the MQTT protocol. A pre-trained deep learning model based on convolutional neural network with attention layer used classify diseases. detect five different categories heartbeats: Normal Beat, Supraventricular premature beat, Premature ventricular contraction, Fusion ventricular, Unclassifiable beat from fever or non-fever Furthermore, provides report patient's rate level, indicating whether are within normal ranges not. automatically connects user nearest doctor further diagnosis if any abnormalities detected.

Language: Английский

---

Using transfer learning-based plant disease classification and detection for sustainable agriculture

BMC Plant Biology, Journal Year: 2024, Volume and Issue: 24(1)

Published: Feb. 26, 2024

Abstract Subsistence farmers and global food security depend on sufficient production, which aligns with the UN's “Zero Hunger,” “Climate Action,” “Responsible Consumption Production” sustainable development goals. In addition to already available methods for early disease detection classification facing overfitting fine feature extraction complexities during training process, how signs of green attacks can be identified or classified remains uncertain. Most pests symptoms are seen in plant leaves fruits, yet their diagnosis by experts laboratory is expensive, tedious, labor-intensive, time-consuming. Notably, diseases appropriately detected timely prevented a hotspot paradigm smart, agriculture unknown. recent years, deep transfer learning has demonstrated tremendous advances recognition accuracy object image systems since these frameworks utilize previously acquired knowledge solve similar problems more effectively quickly. Therefore, this research, we introduce two (PDDNet) models fusion (AE) lead voting ensemble (LVE) integrated nine pre-trained convolutional neural networks (CNNs) fine-tuned efficient identification classification. The experiments were carried out 15 classes popular PlantVillage dataset, 54,305 samples different species 38 categories. Hyperparameter fine-tuning was done models, including DenseNet201, ResNet101, ResNet50, GoogleNet, AlexNet, ResNet18, EfficientNetB7, NASNetMobile, ConvNeXtSmall. We test CNNs stated problem, both independently as part an ensemble. final phase, logistic regression (LR) classifier utilized determine performance various CNN model combinations. A comparative analysis also performed classifiers, learning, proposed model, state-of-the-art studies. that PDDNet-AE PDDNet-LVE achieved 96.74% 97.79%, respectively, compared current when tested several diseases, depicting its exceptional robustness generalization capabilities mitigating concerns

Language: Английский

---

Artificial Neural Network Modeling for the Prediction, Estimation, and Treatment of Diverse Wastewaters: A Comprehensive Review and Future Perspective

Chemosphere, Journal Year: 2024, Volume and Issue: 362, P. 142860 - 142860

Published: July 15, 2024

Language: Английский

---

Wearable Technologies and AI at the Far Edge for Chronic Heart Failure Prevention and Management: A Systematic Review and Prospects

Sensors, Journal Year: 2023, Volume and Issue: 23(15), P. 6896 - 6896

Published: Aug. 3, 2023

Smart wearable devices enable personalized at-home healthcare by unobtrusively collecting patient health data and facilitating the development of intelligent platforms to support care management. The accurate analysis obtained from is crucial for interpreting contextualizing reliable diagnosis management critical chronic diseases. combination edge computing artificial intelligence has provided real-time, time-critical, privacy-preserving solutions. However, based on envisioned service, evaluating additive value overall architecture essential before implementation. This article aims comprehensively analyze current state art smart infrastructures implementing AI technologies at far patients with heart failure (CHF). In particular, we highlight contribution in supporting integration into IoT-aware technology that provide services We also offer an in-depth open challenges potential solutions facilitate innovative technological interactive doctors.

Language: Английский

---

Embedded-machine learning and soft, flexible sensors for wearable devices - viewing from an AI engineer

Materials Today Physics, Journal Year: 2024, Volume and Issue: 42, P. 101376 - 101376

Published: Feb. 23, 2024

Language: Английский

---

Using a novel convolutional neural network for plant pests detection and disease classification

Journal of the Science of Food and Agriculture, Journal Year: 2023, Volume and Issue: 103(12), P. 5849 - 5861

Published: May 13, 2023

Early plant diseases and pests identification reduces social, economic, environmental deficiencies entailing toxic chemical utilization on agricultural farms, thus posing a threat to global food security.An enhanced convolutional neural network (CNN) along with long short-term memory (LSTM) using majority voting ensemble classifier has been proposed tackle pest disease classification. Within pre-trained models, deep feature extractions have obtained from connected layers. Deep features extracted are sent the LSTM layer build robust, LSTM-CNN model for detecting diseases. Experiments were carried out Turkey dataset, 4447 apple categorized into 15 different classes.The study was evaluated in CNNs logistic regression (LR), LSTM, extreme learning machine (ELM), focusing detection problems. The used at detect classify labels. Furthermore, an autonomous selection of optimal parameters applied. Finally, performance validated based sensitivity, F1 score, accuracy, specificity ELM, LR classifiers.The presented attained 99.2% accuracy compared cutting-edge models classifiers such as LR, performed better transfer learning. Pre-trained VGG19, VGG18, AlexNet, demonstrated when fc6 other © 2023 Society Chemical Industry.

Language: Английский

---

A Hybrid Generic Framework for Heart Problem Diagnosis Based on a Machine Learning Paradigm

Sensors, Journal Year: 2023, Volume and Issue: 23(3), P. 1392 - 1392

Published: Jan. 26, 2023

The early, valid prediction of heart problems would minimize life threats and save lives, while lack false diagnosis can be fatal. Addressing a single dataset alone to build machine learning model for the identification is not practical because each country hospital has its own data schema, structure, quality. On this basis, generic framework been built problem diagnosis. This hybrid that employs multiple deep techniques votes best outcome based on novel voting technique with intention remove bias from model. contains two consequent layers. first layer simultaneous models running over given dataset. second consolidates outputs classifies them as classification techniques. Prior process, selects top features using proposed feature selection framework. It starts by filtering columns methods considers common selected. Results framework, 95.6% accuracy, show superiority model, classical stacking technique, traditional technique. main contribution work demonstrate how probabilities exploited purpose creating another final output; step neutralizes any bias. Another experimental proving complete pipeline's ability retrained used other datasets collected different measurements distributions.

Language: Английский

---

CNN and SVM-Based Models for the Detection of Heart Failure Using Electrocardiogram Signals

Sensors, Journal Year: 2022, Volume and Issue: 22(23), P. 9190 - 9190

Published: Nov. 26, 2022

Heart failure (HF) is a serious condition in which the heart fails to supply body with enough oxygen and nutrients function normally. Early accurate detection of critical for impeding disease progression. An electrocardiogram (ECG) test that records rhythm electrical activity used detect HF. It look irregularities heart’s or conduction, as well history attacks, ischemia, other conditions may initiate However, sometimes, it becomes difficult time-consuming interpret ECG signal, even cardiac expert. This paper proposes two models automatically HF from signals: first one introduces Convolutional Neural Network (CNN), while second suggests an extension by integrating Support Vector Machine (SVM) layer classification at end network. The proposed provide more automatic using 2-s fragments. Both are smaller than previously literature when architecture taken into account, reducing both training time memory consumption. MIT-BIH BIDMC databases testing adopted models. experimental results demonstrate effectiveness framework achieving accuracy, sensitivity, specificity over 99% blindfold cross-validation. this study can doctors reliable references be portable devices enable real-time monitoring patients.

Language: Английский

---

Heuristic-based image stitching algorithm with automation of parameters for smart solutions

Expert Systems with Applications, Journal Year: 2023, Volume and Issue: 241, P. 122792 - 122792

Published: Dec. 1, 2023

Language: Английский

---

ConvNeXt-based Mango Leaf Disease Detection: Differentiating Pathogens and Pests for Improved Accuracy

International Journal of Advanced Computer Science and Applications, Journal Year: 2023, Volume and Issue: 14(6)

Published: Jan. 1, 2023

Mango farming is a key economic activity in several locations across the world. trees are prone to various diseases caused by viruses and pests, which can substantially impair crops have an effect on farmers' revenue. To stop spread of these illnesses lessen crop damage they cause, early diagnosis essential. Growing interest has been shown employing deep learning models create automated disease detection systems for because recent developments machine learning. This research article includes study application ConvNeXt pathogen pest mango plants. The intends investigate variety how emerge leaves assess efficiency identifying categorizing them. Images healthy as well with brought pathogens pests included dataset used study. In study, were applied classify pathogens. achieved high accuracy both datasets, better performance dataset. Larger consistently outperformed smaller ones, indicating their ability learn complex features. ConvNeXtXLarge model showed highest accuracy: 98.79% 100% pathogens, 99.17% combined work holds significance detection, aiding efficient management potential benefits farmers. However, models' be influenced quality, preprocessing techniques, hyperparameter selection.

Language: Английский

---