**7. Evolutionary pathway for Network Centric Therapy with respect to quantification of movement disorders, such as Parkinson's disease and Essential tremor**

Functionally wearable accelerometer systems have been demonstrated for the quantification of movement disorder and also their response to intervention strategy [11, 12, 93–98]. With the evolution of wireless technology other traditional inertial signal data transfer strategies have become effectively obsolete [99]. Intuitively, the G-link wireless accelerometer was a candidate for testing and evaluating the quantification of tremor associated with movement disorders [11, 12, 100–103].

Preliminary demonstration of the G-link wireless accelerometer showed the ability to quantify simulated Parkinson's disease hand tremor by mounting the device to the dorsum of the hand [100, 101]. Eventually simulated Parkinson's disease tremor was contrasted to a static condition. Post-processing of the signal data involved the time averaged acceleration, for which statistical significance was achieved [100]. A similar wireless inertial sensor system configuration was successfully demonstrated for the quantification of Parkinson's disease hand tremor within this timeframe [104].

LeMoyne and Mastroianni during 2010 extended the capability of wearable and wireless inertial sensor systems for quantifying Parkinson disease hand tremor through the application of a smartphone. A software application enabled the smartphone to quantify hand tremor for a prescribed temporal duration through the smartphone's internal accelerometer. The accelerometer signal data was conveyed by wireless connectivity to the Internet as an email attachment. Statistical significance was achieved with respect to the subject with Parkinson's disease hand tremor and subject without Parkinson disease. Notably, the experiment occurred in metropolitan Pittsburgh, Pennsylvania and the post-processing was conducted in the general area of Los Angeles, California [105]. The research team observed that experimental and post-processing resources could be geographically separated anywhere in the world with Internet access [1, 2, 5–10, 105, 106]. This observation constitutes the origins of Network Centric Therapy with regards to movement disorders [1, 2, 5–7, 106].

Using the smartphone as an inertial sensor platform with wearable properties the recorded signal data can represent instrumental feedback with respect to the efficacy of therapy response. For example, with machine learning classification the smartphone functioning as a wearable and wireless inertial sensor platform can distinguish between deep brain stimulation set to 'On' and 'Off' status. A person with Essential tremor performed a reach and grasp task with a smartphone mounted to the dorsum of the hand by a latex glove. Post-processing consolidated the inertial signal data to a feature set amenable for machine learning classification, and considerable classification accuracy was achieved through the application of a support vector machine to differentiate between deep brain stimulation set to 'On' and 'Off' status [107]. In conjunction with the preliminary success of the research with respect to Essential tremor and deep brain stimulation set to 'On' and 'Off' status the multilayer perceptron neural network also attained considerable machine learning classification accuracy for differentiating these deep brain stimulation settings [108].

Another extrapolation of this research perspective involved considering six machine learning algorithms: multilayer perceptron neural network, support vector machine, K-nearest neighbors, logistic regression, J48 decision tree, and random forest. The reach and grasp task was applied for a subject with Essential tremor treated by deep brain stimulation with respect to 'On' and 'Off' status. Three feature set scenarios were addressed to determine the most appropriate machine learning

**303**

**Figure 8.**

*brain stimulation intervention [112].*

*An Evolutionary Perspective for Network Centric Therapy through Wearable and Wireless…*

algorithms: accelerometer and gyroscope signal recordings, accelerometer signal recordings, and gyroscope signal recordings. The multilayer perceptron neural network, support vector machine, K-nearest neighbors, and logistic regression achieved the highest classification accuracy in consideration of these three feature

The accelerometer and gyroscope intrinsic to the smartphone was also applied for the evaluation of deep brain stimulation efficacy for the treatment of Parkinson's disease. Deep brain stimulation was set to 'On' and 'Off ' status with the hand tremor response measured by a smartphone mounted to the dorsum of the hand through a latex glove. Multiple machine learning algorithms were evaluated: multilayer perceptron neural network, support vector machine, K-nearest neighbors, logistic regression, J48 decision tree, and random forest. The feature set consisted of descriptive statistics for both the accelerometer and gyroscope signal data. Two performance parameters were considered, such as classification accuracy and time to develop the machine learning model. The support vector machine and logistic regression best satisfied these two performance parameters [110]. The multilayer perceptron neural network achieved considerable classification accuracy to distinguish between the deep brain stimulation set to 'On' and 'Off' status for Parkinson's disease hand tremor, but the time to develop the model was

Network Centric Therapy was further realized for the domain of movement disorders through the BioStamp nPoint. The BioStamp nPoint is a conformal wearable and wireless inertial sensor system with segmented operation and wireless transmission of signal data to a secure Cloud computing environment with wireless connectivity to a smartphone and tablet. The conformal sensors also have a mass less than ten grams and a profile on the order of a bandage. Additionally, the BioStamp nPoint is certified as an FDA 510(k) medical device for the acquisition of medical grade data [5, 90]. These attributes of the BioStamp nPoint ideally accommodate the quantification of movement disorder tremor response, such as for Parkinson's disease, based on deep brain stimulation intervention through mounting about the dorsum of the hand using an adhesive medium as illustrated in **Figure 8** [112]. Multiple sets of deep brain stimulation parameter configurations have been evaluated for the treatment of Parkinson's disease using the BioStamp nPoint to quantify the response and machine learning to distinguish the respective parameter configurations [112–115]. The BioStamp nPoint was mounted to the dorsum of

*The BioStamp nPoint conformal wearable and wireless inertial sensor system mounted about the dorsum of the hand for quantifying movement disorder tremor response, such as for Parkinson's disease, as a result of deep* 

*DOI: http://dx.doi.org/10.5772/intechopen.95550*

considerably protracted [110, 111].

set scenarios [109].

#### *An Evolutionary Perspective for Network Centric Therapy through Wearable and Wireless… DOI: http://dx.doi.org/10.5772/intechopen.95550*

algorithms: accelerometer and gyroscope signal recordings, accelerometer signal recordings, and gyroscope signal recordings. The multilayer perceptron neural network, support vector machine, K-nearest neighbors, and logistic regression achieved the highest classification accuracy in consideration of these three feature set scenarios [109].

The accelerometer and gyroscope intrinsic to the smartphone was also applied for the evaluation of deep brain stimulation efficacy for the treatment of Parkinson's disease. Deep brain stimulation was set to 'On' and 'Off ' status with the hand tremor response measured by a smartphone mounted to the dorsum of the hand through a latex glove. Multiple machine learning algorithms were evaluated: multilayer perceptron neural network, support vector machine, K-nearest neighbors, logistic regression, J48 decision tree, and random forest. The feature set consisted of descriptive statistics for both the accelerometer and gyroscope signal data. Two performance parameters were considered, such as classification accuracy and time to develop the machine learning model. The support vector machine and logistic regression best satisfied these two performance parameters [110]. The multilayer perceptron neural network achieved considerable classification accuracy to distinguish between the deep brain stimulation set to 'On' and 'Off' status for Parkinson's disease hand tremor, but the time to develop the model was considerably protracted [110, 111].

Network Centric Therapy was further realized for the domain of movement disorders through the BioStamp nPoint. The BioStamp nPoint is a conformal wearable and wireless inertial sensor system with segmented operation and wireless transmission of signal data to a secure Cloud computing environment with wireless connectivity to a smartphone and tablet. The conformal sensors also have a mass less than ten grams and a profile on the order of a bandage. Additionally, the BioStamp nPoint is certified as an FDA 510(k) medical device for the acquisition of medical grade data [5, 90]. These attributes of the BioStamp nPoint ideally accommodate the quantification of movement disorder tremor response, such as for Parkinson's disease, based on deep brain stimulation intervention through mounting about the dorsum of the hand using an adhesive medium as illustrated in **Figure 8** [112].

Multiple sets of deep brain stimulation parameter configurations have been evaluated for the treatment of Parkinson's disease using the BioStamp nPoint to quantify the response and machine learning to distinguish the respective parameter configurations [112–115]. The BioStamp nPoint was mounted to the dorsum of

#### **Figure 8.**

*Wireless Sensor Networks - Design, Deployment and Applications*

**Essential tremor**

this timeframe [104].

disorders [1, 2, 5–7, 106].

**7. Evolutionary pathway for Network Centric Therapy with respect to quantification of movement disorders, such as Parkinson's disease and** 

Functionally wearable accelerometer systems have been demonstrated for the quantification of movement disorder and also their response to intervention strategy [11, 12, 93–98]. With the evolution of wireless technology other traditional inertial signal data transfer strategies have become effectively obsolete [99]. Intuitively, the G-link wireless accelerometer was a candidate for testing and evaluating the quanti-

Preliminary demonstration of the G-link wireless accelerometer showed the ability to quantify simulated Parkinson's disease hand tremor by mounting the device to the dorsum of the hand [100, 101]. Eventually simulated Parkinson's disease tremor was contrasted to a static condition. Post-processing of the signal data involved the time averaged acceleration, for which statistical significance was achieved [100]. A similar wireless inertial sensor system configuration was successfully demonstrated for the quantification of Parkinson's disease hand tremor within

LeMoyne and Mastroianni during 2010 extended the capability of wearable and

Using the smartphone as an inertial sensor platform with wearable properties the recorded signal data can represent instrumental feedback with respect to the efficacy of therapy response. For example, with machine learning classification the smartphone functioning as a wearable and wireless inertial sensor platform can distinguish between deep brain stimulation set to 'On' and 'Off' status. A person with Essential tremor performed a reach and grasp task with a smartphone mounted to the dorsum of the hand by a latex glove. Post-processing consolidated the inertial signal data to a feature set amenable for machine learning classification, and considerable classification accuracy was achieved through the application of a support vector machine to differentiate between deep brain stimulation set to 'On' and 'Off' status [107]. In conjunction with the preliminary success of the research with respect to Essential tremor and deep brain stimulation set to 'On' and 'Off' status the multilayer perceptron neural network also attained considerable machine learning classification accuracy for differentiating these deep brain stimulation

Another extrapolation of this research perspective involved considering six machine learning algorithms: multilayer perceptron neural network, support vector machine, K-nearest neighbors, logistic regression, J48 decision tree, and random forest. The reach and grasp task was applied for a subject with Essential tremor treated by deep brain stimulation with respect to 'On' and 'Off' status. Three feature set scenarios were addressed to determine the most appropriate machine learning

wireless inertial sensor systems for quantifying Parkinson disease hand tremor through the application of a smartphone. A software application enabled the smartphone to quantify hand tremor for a prescribed temporal duration through the smartphone's internal accelerometer. The accelerometer signal data was conveyed by wireless connectivity to the Internet as an email attachment. Statistical significance was achieved with respect to the subject with Parkinson's disease hand tremor and subject without Parkinson disease. Notably, the experiment occurred in metropolitan Pittsburgh, Pennsylvania and the post-processing was conducted in the general area of Los Angeles, California [105]. The research team observed that experimental and post-processing resources could be geographically separated anywhere in the world with Internet access [1, 2, 5–10, 105, 106]. This observation constitutes the origins of Network Centric Therapy with regards to movement

fication of tremor associated with movement disorders [11, 12, 100–103].

**302**

settings [108].

*The BioStamp nPoint conformal wearable and wireless inertial sensor system mounted about the dorsum of the hand for quantifying movement disorder tremor response, such as for Parkinson's disease, as a result of deep brain stimulation intervention [112].*

the hand through an adhesive medium. The deep brain stimulation amplitude was evaluated at multiple settings, such as 'Off' status as a baseline, amplitude set to 1.0 mA, 2.5 mA, and 4.0 mA. The acceleration signal derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system was post-processed to present the acceleration magnitude as illustrated in **Figures 9**–**12** [112].

#### **Figure 9.**

*Acceleration magnitude derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system for hand tremor from a subject with Parkinson's disease with deep brain stimulation set to 'Off' status [112].*

#### **Figure 10.**

*Acceleration magnitude derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system for hand tremor from a subject with Parkinson's disease with deep brain stimulation set to amplitude equal to 1.0 mA [112].*

**305**

**Figure 12.**

*equal to 4.0 mA [112].*

**Figure 11.**

*equal to 2.5 mA [112].*

*An Evolutionary Perspective for Network Centric Therapy through Wearable and Wireless…*

*Acceleration magnitude derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system for hand tremor from a subject with Parkinson's disease with deep brain stimulation set to amplitude* 

The acceleration magnitude signal data was consolidated to a feature set though Python. The feature set was composed of numeric attributes, such as maximum, minimum, mean, standard deviation, and coefficient of variation. Machine learning algorithms, such as J48 decision tree, K-nearest neighbors, support vector machine, logistic regression, and random forest were contrasted in terms of their

*Acceleration magnitude derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system for hand tremor from a subject with Parkinson's disease with deep brain stimulation set to amplitude* 

*DOI: http://dx.doi.org/10.5772/intechopen.95550*

*An Evolutionary Perspective for Network Centric Therapy through Wearable and Wireless… DOI: http://dx.doi.org/10.5772/intechopen.95550*

**Figure 11.**

*Wireless Sensor Networks - Design, Deployment and Applications*

the hand through an adhesive medium. The deep brain stimulation amplitude was evaluated at multiple settings, such as 'Off' status as a baseline, amplitude set to 1.0 mA, 2.5 mA, and 4.0 mA. The acceleration signal derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system was post-processed

*Acceleration magnitude derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system for hand tremor from a subject with Parkinson's disease with deep brain stimulation set to 'Off'* 

*Acceleration magnitude derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system for hand tremor from a subject with Parkinson's disease with deep brain stimulation set to amplitude* 

to present the acceleration magnitude as illustrated in **Figures 9**–**12** [112].

**304**

**Figure 10.**

*equal to 1.0 mA [112].*

**Figure 9.**

*status [112].*

*Acceleration magnitude derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system for hand tremor from a subject with Parkinson's disease with deep brain stimulation set to amplitude equal to 2.5 mA [112].*

#### **Figure 12.**

*Acceleration magnitude derived from the BioStamp nPoint conformal wearable and wireless inertial sensor system for hand tremor from a subject with Parkinson's disease with deep brain stimulation set to amplitude equal to 4.0 mA [112].*

The acceleration magnitude signal data was consolidated to a feature set though Python. The feature set was composed of numeric attributes, such as maximum, minimum, mean, standard deviation, and coefficient of variation. Machine learning algorithms, such as J48 decision tree, K-nearest neighbors, support vector machine, logistic regression, and random forest were contrasted in terms of their

classification accuracy and time to develop the machine learning model. Based on these criteria the K-nearest neighbors machine learning algorithm displayed the optimal satisfaction of classification accuracy in conjunction with time to develop the machine learning model and the support vector machine achieved the optimal classification accuracy [112]. The multilayer perceptron neural network also demonstrated considerable classification accuracy [113].

Deep learning was then applied to distinguish between deep brain stimulation parameter configuration settings for the treatment of Parkinson's disease, such as 'Off' status as a baseline, amplitude set to 1.0 mA, amplitude set to 1.75 mA, amplitude set to 2.5 mA, amplitude set to 3.25 mA, and amplitude set to 4.0 mA. The BioStamp nPoint conformal wearable and wireless inertial sensor system provided the accelerometer signal data. The post-processing was facilitated by Google Colab and TensorFlow to implement a convolutional neural network. The convolutional neural network achieved considerable classification accuracy to distinguish between all six of these parameter configurations [116, 117].
