Table 11.
IoT/IoMT-based detection of different diseases.
| Wearable/Smart/Medical Device | Machine Learning Approach | Results |
|---|---|---|
| Smartphone Samsung Galaxy Young (Samsung Electronics Co., Yeongtong-gu, Suwon-si, South Korea.) [111] |
SVM | Accuracy: 97.74%, Precision: 92.21% |
| EEG sensors, ECG sensors, accelerometer, gateway module [112] |
Random Forest | Accuracy: 83.35%, Precision: 91.32% Recall: 91.32%, F1-Score: 65% |
| SmartCardia INYU, (SmartCardia Inc., Lausanne, Switzerland) [113] | Random Forest | Sensitivity: 87.95%, Specificity: 78.82% |
| Electronic nose: 1 humidity sensor 18 electrochemical gas sensors [114] |
SVM | Accuracy: 97.19%, Sensitivity: 93.37% Specificity: 99.07% |
| Network on body-area sensor (BAS) Raspberry Pi 3B+, (Raspberry Pi Foundation, Cambridge, UK) [115] |
Deep Neural Network (DNN) | Accuracy: 90% |
| Smart device sensors [116] | ResNet-9, federated semi-supervised learning (FSSL) |
Accuracy: 95.9% |
| Photoplethysmography sensor, temperature sensor, accelerometer 12C slave sensor, microcontroller 12C master, (Custom-made, Miami, USA) [117] |
Long Short-Term Memory (LSTM) | Root mean square: 0.07% Accuracy: 99.5% |
| Motion sensor, ECG sensor EMG sensor, Foot sensor [118] |
Long Short-Term Memory (LSTM) | Accuracy: 98.99% |
| Raspberry Pi 3B, (Raspberry Pi Foundation, Cambridge, UK) NVidia Jetson Nano, (NVidia, Santa Clara, CA, USA) [119] |
Deep Neural Network (DNN) | Accuracy: 99.8% |
| ECG sensors [120] | R-peak detection algorithm | Reduction of the data dropout rate, by average of 21.09% Number of R-peak detections increased by 15.33% compared to the existing classification system |
| NRF52 cortex ARM M4F microcontroller (NRF52DK), (Nordic semiconductor, Trondheim, Norway) [121] |
Artificial Neural Network (ANN) | INCART Accuracy: 93% INCART Sensitivity: 88% INCART Specificity: 94% INCART Precision: 67% |