Otwarty dostęp

Investigation of Deep Learning Models for Analysis of Heart Disorders in Smart Health Care based IoT Environment

Jewel Sengupta

Jewel Sengupta

Kaunas University of TechnologyKaunas, Lithuania
Wyszukaj tego autora
Sciendo|Google Scholar
Sengupta, Jewel
  
15 cze 2024
Investigation of Deep Learning Models for Analysis of Heart Disorders in Smart Health Care based IoT Environment's Cover Image
O artykule
Poprzedni artykuł
Następny artykuł
Zacytuj
Pobierz okładkę
Kategoria artykułu: Article
Data publikacji: 15 cze 2024
Zakres stron: 1 - 16
Otrzymano: 22 mar 2024
Przyjęty: 11 kwi 2024
DOI: https://doi.org/10.2478/jsiot-2024-0001
Słowa kluczowe
© 2023 Jewel Sengupta, published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.

Figure 1:

Proposed Framework
Proposed Framework

Figure 2:

GRU -network Architecture.
GRU -network Architecture.

Figure 3:

ROC curves a)UCI datasets b) Framingham c) Public Datasets c) real time Sensor inputs.
ROC curves a)UCI datasets b) Framingham c) Public Datasets c) real time Sensor inputs.

Figure 4:

Evaluation metrics of Proposed Model utilising the UCI Datasets
Evaluation metrics of Proposed Model utilising the UCI Datasets

Figure 5:

Performance metrics of Proposed Model using the Firmangham Datasets
Performance metrics of Proposed Model using the Firmangham Datasets

Figure 6:

Performance metrics of Proposed Model using the Public Health Datasets
Performance metrics of Proposed Model using the Public Health Datasets

Figure 7:

Performance metrics of Proposed Model utilising the Real time Sensor Datasets
Performance metrics of Proposed Model utilising the Real time Sensor Datasets

Real-Time Data Used for the Testing and Evaluation

Dataset Description Dataset Description No. of Records No. of Attributes Associated Tasks Training Data / Testing
Real-Time Datasets 1,190 200 14 Classification 80:20

Comparative Analysis of the Different Algorithm In Handling the Public health datasets

Algorithm Accuracy (%) Precision (%) Recall (%) Specificity (%) F1-Score (%)
LSTM 91.3% 90.4% 90.2% 89.2% 89.3%
CNN+LSTM 91.5% 91.34% 91.48% 91.35% 91.6%
RNN+LSTM 92.4% 93.7% 93.0% 93.2% 93.0%
HRFLM 89.0% 89.9% 89.78% 89.68% 90%
RFRS 86.4% 87.2% 87.4% 87.43% 87.2%
MDCNN 95.2% 95.2% 94.9% 94.5% 94.35%
Proposed Model 98.17% 98.1% 98.1% 98.17% 98.15%

Comparative Analysis of the Different Algorithm In Handling the Framingham datasets

Algorithm Accuracy (%) Precision (%) Recall (%) Specificity (%) F1-Score (%)
LSTM 91.3% 90.4% 90.2% 89.2% 89.3%
CNN+LSTM 91.5% 91.34% 91.48% 91.35% 91.6%
RNN+LSTM 92.4% 93.7% 93.0% 93.2% 93.0%
HRFLM 89.0% 89.9% 89.78% 89.68% 90%
RFRS 86.4% 87.2% 87.4% 87.43% 87.2%
MDCNN 95.2% 95.2% 94.9% 94.5% 94.35%
Proposed Model 98.17% 98.1% 98.1% 98.17% 98.15%

Evaluation Metrics utilized for the assessment

SL.NO Evaluation Metrics Mathematical Expression
01 Accuracy TP+TNTP+TN+FP+FN {{TP + TN} \over {TP + TN + FP + FN}}
02 Recall TPTP+FN×100 {{{\rm{TP}}} \over {{\rm{TP}} + {\rm{FN}}}} \times 100
03 Specificity TNTN+FP {{TN} \over {TN + FP}}
04 Precision TNTP+FP {{TN} \over {TP + FP}}
05 F1-Score 2.Precison ∗ RecallPrecison + Recall 2.{{Precison\, * \,{Recall}} \over {Precision\, + \, {Recall}}}

Comparative Assessment of Distinct Algorithm In Handling the UCI datasets

Algorithm Accuracy (%) Precision (%) Recall (%) Specificity (%) F1-Score (%)
LSTM 91.3% 90.4% 90.2% 89.2% 89.3%
CNN+LSTM 91.5% 91.34% 91.48% 91.35% 91.6%
RNN+LSTM 92.4% 93.7% 93.0% 93.2% 93.0%
HRFLM 89.0% 89.9% 89.78% 89.68% 90%
RFRS 86.4% 87.2% 87.4% 87.43% 87.2%
MDCNN 95.2% 95.2% 94.9% 94.5% 94.35%
Proposed Model 98.17% 98.1% 98.1% 98.17% 98.15%

Comparative Analysis of the Different Algorithm In Handling the Real time Datasets

Algorithm Accuracy (%) Precision (%) Recall (%) Specificity (%) F1-Score (%)
LSTM 91.3% 90.4% 90.2% 89.2% 89.3%
CNN+LSTM 91.5% 91.34% 91.48% 91.35% 91.6%
RNN+LSTM 92.4% 93.7% 93.0% 93.2% 93.0%
HRFLM 89.0% 89.9% 89.78% 89.68% 90%
RFRS 86.4% 87.2%% 87.4% 87.43% 87.2%
MDCNN 95.2% 95.2% 94.9% 94.5% 94.35%
Proposed Model 98.17% 98.1% 98.1% 98.17% 98.15%

Datasets Details Used for the Experimentation

Dataset Description Dataset Description No. of Records No. of Attributes Associated Tasks Training Data / Testing
UCI Machine 18,000 203 55 Classification 80:20
Learning Public Health 3,300 1,000 9 Prediction 80:20
Datasets Framingham 3,780 3800 10 Prediction 80:20
Język:
Angielski
Częstotliwość wydawania:
2 razy w roku
Dziedziny czasopisma:
Inżynieria, Wstępy i przeglądy, Historia inżynierii, Elektrotechnika, Podstawy elektrotechniki, Elektronika, Technologia informacyjna
Kanał RSS czasopisma