Table 9.

IoT/IoMT-based blood pressure and hypertension detection.

Wearable/Smart/Medical Device	Machine Learning Approach	Results
Accelerometer, GPS, ECG, Blood Pressure Monitor [78]	Multilayer Perceptron, Decision Tree J48, Decision Table, Radial Basis Function, Bayes Network	Accuracy for three different types of patient Accuracy MLP: 91.46%, 95.54%, 90.75%, Accuracy J48: 99.14%, 99.78%, 99.1%, Accuracy DTable: 95.91%, 97.08%, 96.33%, Accuracy RBF: 81.52%, 83.95%, 84.78%, Accuracy BN: 86.58%, 95.11%, 88.55%
Impedance cardiography sensor, (Custom-made, India) [79]	Auto-adaptive algorithm based on Impedance Cardiography signals for non-invasive, cuffless, continous monitoring of blood pressure and heart rate	Systolic BP: ±2.33 mmHg Diastolic BP: ±3.60 mmHg Heart rate: ±2.88 beats
ADS1299EEG-FE, (TX Instruments, Dallas, TX, USA), AFE4490SPO2, (TX Instruments, Dallas, TX, USA), MSP430F55291PN, (TX Instruments, Dallas, TX, USA) [80]	SVM, Dynamic Time Warping (DTW), K-medoids clustering	ME ${}^1\pm$ STD ${}^2$: $0.8\pm 2.7$ BPM ${}^3$, MAE ${}^4$: 1.8 BPM, RMSE ${}^5$: 2.8 BPM For HR estimation
Ring PPG, Accelerometer, ZigBee, (Custom-made device, Taiwan) [81]	MIL (Multiplate instance learning algorithm)	Accuracy Standard Deviation of all RR (NN) intervals: 85.74%, Specificity: 83.33%, Precision: 92.11%, Sensitivity: 86.42%
Raspberry Pi 2, (Raspberry Pi Foundation, Cambridge, UK) [82]	Random Forest, Decision Tree, SVM, AdaBoost	SBP ${}^6$ RMSE ${}^5$: $3.2\pm 0.7$ mmHg, DBP ${}^7$ RMSE ${}^5$: $2.2\pm 0.7$ mmHg, SBP ${}^6$ MAE ${}^4$: $4.4\pm 1.0$ mmHg, DBP ${}^7$ MAE ${}^4$: $2.9\pm 1.2$ mmHg
Pulse oximeter, (Arduino, Scarmagno, Italy) [83]	k-NN, SVM, Decision Tree, Neural Network	10 fold cross k-NN Precision: 91%, SVM Precision: 96%, DT Precision: 95%, NN Precision: 96%, LOOCV ${}^8$ k-NN Precision: 90%, SVM Precision: 93%, DT Precision: 94%, NN Precision: 95%
CMS50FW Pulse Oximeter, (Contec Inc., Qinhuangdao, China) Finometer MIDI Model II, (Finapres Medical Systems B.V., Amsterdam, The Netherlands) [84]	SVM	MAE ${}^4$: systolic 0.043 mmHg, diastolic 0.011 mmHg, mean blood pressure 0.008 mmHg
Mindray N12, (Mindray, Shenzhen, China) [85]	Residual Network Long Short-Term Memory Network (Res-LSTM)	SBP${}^6$ Mean difference ± Standard deviation accuracy: −0.2 ± 5.82 mmHg, Mean Arterial Pressure Mean difference ± Standard deviation accuracy: −0.57 ± 4.39 mmHg DBP ${}^7$, Mean difference ± Standard deviation accuracy: −0.75 ± 5.62 mmHg

¹ Mean error. ² Standard deviation. ³ Beats per minute. ⁴ Mean absolute error. ⁵ Root mean square error. ⁶ Systolic blood pressure. ⁷ Diastolic blood pressure. ⁸ Leave-one-out cross-validation.