A decision-making mechanism for context inference in pervasive healthcare environments
Decision Support Systems
| Hi-index | 0.00 |
Objective: The objective of this research study is to derive a comprehensive physiological model relating to subjects under physical stress conditions. The model should describe the behaviour of the cardiovascular system, respiratory system, thermoregulation and brain activity in response to physical workload. Methods and material: An experimental testing rig was built which consists of recumbent high performance bicycle for inducing the physical load and a data acquisition system comprising monitors and PCs. The signals acquired and used within this study are the blood pressure, heart rate, respiration, body temperature, and EEG signals. The proposed model is based on a grey-box based modelling approach which was used because of the sufficient level of details it provides. Cardiovascular and EEG Data relating to 16 healthy subject volunteers (data from 12 subjects were used for training/validation and the data from 4 subjects were used for model testing) were collected using the Finapres^(R) and the ProComp+^(R) monitors. For model validation, residual analysis via the computing of the confidence intervals as well as related histograms was performed. Results: Closed-loop simulations for different subjects showed that the model can provide reliable predictions for heart rate, blood pressure, body temperature, respiration, and the EEG signals. These findings were also reinforced by the residual analyses data obtained, which suggested that the residuals were within the 90% confidence bands and that the corresponding histograms were of a normal distribution. A higher intelligent level was added to the model, based on neural networks, to extend the capabilities of the model to predict over a wide range of subjects dynamics. Conclusion: The elicited physiological model describing the effect of physiological stress on several physiological variables can be used to predict performance breakdown of operators in critical environments. Such a model architecture lends itself naturally to exploitation via feedback control in a 'reverse-engineering' fashion to control stress via the specification of a safe operating range for the psycho-physiological variables.