Review Article

Validation Techniques for Sensor Data in Mobile Health Applications

Table 1

Classification of the data validation methods by functionality.

Groups of data validation methodsMethods includedDescription

Faulty data detection methodsANNs
 (i) MLP; AANN; BP algorithm; SVM;
Instance based
 (i) SOM
Gaussian distributions
Statistical methods
 (i) ASV; HSV
Probabilistic methods
 (i) Bayesian Networks; Propagation in Trees; Probabilistic Causal Methods; Learning Algorithms; Sparse Bayesian Learning; RVM; SPRT
Dimensionality Reduction
 (i) Fuzzy logic; PCA; KPCA;
others
 (i) Hybrid AANN-KPCA
Consisting of the detection of faulty or incorrect values discovered during the data acquisition and processing stages

Data correction methodsKalman filter
LPC
ARMA
 (i) AR; MA; EMD
Nadaraya-Watson statistical estimator
Interpolation
Smoothing
Data mining techniques
Data reconciliation techniques
Consisting of the estimation of faulty or incorrect values obtained during the data acquisition and processing stages

Other assisting techniques or toolsChecking of the status of the sensors
Checking of the duration after sensor maintenance
Data context classification
Calibration of measuring systems
Uncertainty consideration
Grey models
 (i) GBM; dynamic uncertainty estimation of self-validating sensor
VRFV method
These are different approaches created for the correct validation of the data