**6. Experimental results**

In this chapter, we use the features that we get from Chapter 4 to perform our experiment. There are three experiments: classification with support vector machine, prediction with adaptive neural fuzzy inference systems and cooperating with radar system.

In the first experiment, we choose one feature from *Dp* ´ , *Vp* ´ and the other feature is *γ<sup>V</sup>* . There are two combinations. Among these two combinations, we find the best one according to the classification results and it becomes the inputs of the prediction experiment. In the second experiment, we use the features that we find in the first experiment as the ANFIS inputs and calculate the correlation, MSE, regression slop under different membership functions. In the third experiment, we combine the features of radar system with our features to perform classification and prediction again.

#### **6.1. Classification with support vector machine**

Support vector machine is one of the most widely used machine-learning algorithms for classification problems [22].

We group the subjects of level 1 and level 2 into the Bad group, and the subjects of level 3 belong to the Good group. There are 32 subjects in the Bad group and 28 subjects in the Good group. Those subjects who belong to the Good group are marked with blue triangles and those subjects who belong to the Bad group are marked with red circles.


**Table 2.** The SVM accuracy in different *S*-value with *σD*(1,*<sup>S</sup>* ) ´ and *σ<sup>V</sup>* (1,*<sup>S</sup>* ) ´ .

According to the above chapter, we divide all steps into *S* sections. In order to find the *S*-value, we perform different *S*-values in SVM classification. The inputs are *σD*(1,*<sup>S</sup>* ) ´ and *σ<sup>V</sup>* (1,*<sup>S</sup>* ) ´ because these two parameters are affected by the *S*-value. **Table 2** lists the SVM accuracy in different *S*-values and the highest accuracy is **0.61** when *S* equals six. Therefore, the *S*-value is six in our experiment. In this article, the bold values in all the tables means the best result in the experiment.


**Table 3.** The *Dp* ´ and *Vp* ´ values. Classifying and Predicting Respiratory Function Based on Gait Analysis http://dx.doi.org/10.5772/63917 17

**Figure 15.** The *Dp* ´ and *Vp* ´ values of all subjects.

**6. Experimental results**

16 Advanced Biosignal Processing and Diagnostic Methods

classification and prediction again.

classification problems [22].

experiment.

**Table 3.** The *Dp*

Level *Dp*

´ values.

´ and *Vp*

**6.1. Classification with support vector machine**

subjects who belong to the Bad group are marked with red circles.

*Level* 1 47.72 117.4 *Level* 2 50.49 129.2 *Level* 3 **56.55 152.3**

**Table 2.** The SVM accuracy in different *S*-value with *σD*(1,*<sup>S</sup>* ) ´ and *σ<sup>V</sup>* (1,*<sup>S</sup>* ) ´ .

In this chapter, we use the features that we get from Chapter 4 to perform our experiment. There are three experiments: classification with support vector machine, prediction with

are two combinations. Among these two combinations, we find the best one according to the classification results and it becomes the inputs of the prediction experiment. In the second experiment, we use the features that we find in the first experiment as the ANFIS inputs and calculate the correlation, MSE, regression slop under different membership functions. In the third experiment, we combine the features of radar system with our features to perform

Support vector machine is one of the most widely used machine-learning algorithms for

We group the subjects of level 1 and level 2 into the Bad group, and the subjects of level 3 belong to the Good group. There are 32 subjects in the Bad group and 28 subjects in the Good group. Those subjects who belong to the Good group are marked with blue triangles and those

According to the above chapter, we divide all steps into *S* sections. In order to find the *S*-value, we perform different *S*-values in SVM classification. The inputs are *σD*(1,*<sup>S</sup>* ) ´ and *σ<sup>V</sup>* (1,*<sup>S</sup>* ) ´ because these two parameters are affected by the *S*-value. **Table 2** lists the SVM accuracy in different *S*-values and the highest accuracy is **0.61** when *S* equals six. Therefore, the *S*-value is six in our experiment. In this article, the bold values in all the tables means the best result in the

´ *Vp*

´

*S*-value 1 2 3 4 5 6 7 Accuracy 0.55 0.55 0.55 0.56 0.56 **0.61** 0.58

´ , *Vp*

´ and the other feature is *γ<sup>V</sup>* . There

adaptive neural fuzzy inference systems and cooperating with radar system.

In the first experiment, we choose one feature from *Dp*

From **Table 3**, the subjects of level 3 have larger *Dp* ´ and *Vp* ´ than that of levels 2 and 1. The values of each level are the mean of *Dp* ´ and *Vp* ´ of the subjects who belong to the level. Therefore, the *Dp* ´ and *Vp* ´ become our choices of input features. **Figure 15** shows the *Dp* ´ and *Vp* ´ of all subjects.


**Table 4.** The *σD*(1,6) ´ and *σ<sup>V</sup>* (1,6) ´ values in three levels.

**Figure 16.** The *σD*(1,6) ´ and *σ<sup>V</sup>* (1,6) ´ values of all subjects.

The subjects who have better respiratory function have larger *σD*(1,6) ´ and *σ<sup>V</sup>* (1,6) ´ than those who have poor respiratory function. In **Table 4**, the values of level 3 are higher than those of levels 1 and 2. The values of each level are the mean of *σD*(1,6) ´ and *σ<sup>V</sup>* (1,6) ´ of the subjects who belong to the level. **Figure 16** shows the *σD*(1,6) ´ and *σ<sup>V</sup>* (1,6) ´ of all subjects.

In addition, people have lower variance after they exercise. **Figure 17** shows the pace distance of a person in three different conditions: normal walking, walking after a short run and walking after a long run. The variation of walking after a long run is smaller than others.

**Figure 17.** Pace distance of a person in three conditions.

The *σ<sup>V</sup>* (5,6) ´ is the mean of velocity variance of the hind of the test and *σ<sup>V</sup>* (1,2) ´ is the mean of velocity variance of the front of the test. To those subjects who have poor respiratory function, the *σ<sup>V</sup>* (5,6) ´ is smaller than *σ<sup>V</sup>* (1,2) ´ because they feel like walking after a long run in the hind of the test. On the other hand, for those subjects who have better respiratory function feel like normal walking in the hind of the test.

Therefore, the *σ<sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ of those subjects who have poor respiratory function should be smaller than those subjects who have better respiratory function. From **Table 5**, the *<sup>σ</sup><sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ value of level 3 is greater than levels 1 and 2. The values of each level are the mean of *σ<sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ of the subjects who belong to the level. **Figure 18** shows the *σ<sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ of all subjects.


**Table 5.** The *σ<sup>V</sup>* (5,6) ´ *<sup>σ</sup><sup>V</sup>* (1,2) ´ values in three levels.

**Figure 18.** The *σ<sup>V</sup>* (5,6) ´ *<sup>σ</sup><sup>V</sup>* (1,2) ´ of all subjects.

Classifying and Predicting Respiratory Function Based on Gait Analysis http://dx.doi.org/10.5772/63917 19

**Figure 19.** The *γV* of all subjects.

**Figure 17.** Pace distance of a person in three conditions.

18 Advanced Biosignal Processing and Diagnostic Methods

normal walking in the hind of the test.

Level *<sup>σ</sup><sup>V</sup>* (5,6) ´

*Level* 1 0.93 *Level* 2 0.96 *Level* 3 **1.04**

*<sup>σ</sup><sup>V</sup>* (1,2) ´ values in three levels.

*<sup>σ</sup><sup>V</sup>* (1,2) ´ of all subjects.

subjects.

**Table 5.** The *σ<sup>V</sup>* (5,6) ´

**Figure 18.** The *σ<sup>V</sup>* (5,6) ´

The *σ<sup>V</sup>* (5,6) ´ is the mean of velocity variance of the hind of the test and *σ<sup>V</sup>* (1,2) ´ is the mean of velocity variance of the front of the test. To those subjects who have poor respiratory function, the *σ<sup>V</sup>* (5,6) ´ is smaller than *σ<sup>V</sup>* (1,2) ´ because they feel like walking after a long run in the hind of the test. On the other hand, for those subjects who have better respiratory function feel like

Therefore, the *σ<sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ of those subjects who have poor respiratory function should be smaller than those subjects who have better respiratory function. From **Table 5**, the *<sup>σ</sup><sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ value of level 3 is greater than levels 1 and 2. The values of each level are the mean of *σ<sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ of the subjects who belong to the level. **Figure 18** shows the *σ<sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ of all

*<sup>σ</sup><sup>V</sup>* (1,2) ´


**Table 6.** The accuracy of SVM with different features.

The *γV* considers the mean of velocity variance value and the velocity variance ratio (*γ<sup>V</sup>* <sup>=</sup>*σ<sup>V</sup>* (1,6) ´ \*(*σ<sup>V</sup>* (5,6) ´ / *<sup>σ</sup><sup>V</sup>* (1,2) ´ )), so *γV* become our choices of the input features. **Figure 19** shows *γV* values of all subjects. **Table 6** lists the SVM results with different features.

The accuracy of input features (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ) is better than the other input features (*Dp* ´ , *<sup>γ</sup><sup>V</sup>* ). There‐ fore, we use (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ) as the best inputs of the classification experiment. **Figures 20** and **<sup>21</sup>** show the SVM result with the inputs (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ) and (*Dp* ´ , *<sup>γ</sup><sup>V</sup>* ), respectively. In **Figure 20**, the subjects of the Good group have higher *Vp* ´ and *<sup>γ</sup>V* than those who are in the Bad group.

**Figure 20.** SVM classification result with *Vp* ´ and *γ<sup>V</sup>* .

**Figure 21.** SVM classification result with *Dp* ´ and *γ<sup>V</sup>* .

#### **6.2. Prediction with adaptive neural fuzzy inference systems**

We utilize adaptive neural fuzzy inference system to help us predict the parameters from pulmonary spirometer. The ANFIS system comes from the toolbox of Matlab.

Because we only collect about 60 cases so far, it is not enough for ANFIS to perform prediction. In order to increase the training samples, we adopt **Leave-one-out cross-validation** method. **Leave-one-out cross-validation** is used in analysing small datasets. It uses one sample as the validation set and the remaining as the training set. Repeat on this way for all samples. We can use this method to solve the insufficient data problem.

In ANFIS, it is important to choose a correct membership function. In addition, we also need to choose the input sections. In the experiment, we use six different membership functions including trapmf, gbellmf, gaussmf, gauss2mf, pimf and dsigmf. **Figure 22** shows the mem‐ bership functions we use in our prediction experiment. The inputs of the experiment are *Vp* ´ and *γ<sup>V</sup>* . In our results, we show the correlation, normalized Mean Square Error (MSE*N*) and regression slope under different membership functions. The formula of MSE*<sup>N</sup>* is shown in Eq. (20). The Target*<sup>i</sup>* are the measured values from pulmonary spirometer and the Predict*<sup>i</sup>* are the values come from ANFIS.

$$MSE\_N = \frac{1}{n} \sum\_{l=1}^{n} \left(\frac{Target\_l - Predicted\_l}{Mean(Target)}\right)^2 \tag{20}$$

We try to predict three different parameters that come from the pulmonary spirometer: post FEV1 *FVC* , postFEV1 and postFVC. The 'post' name means the parameters after the 6-min brisk walking test. In the following part, for the convenience, we call post FEV1 FVC , postFEV1, postFVC as, FEV1 FVC FEV1, FVC, respectively. FEV1 FVC and FEV1 are used to access the respiratory function, so we choose these two parameters as our predicting targets.

Classifying and Predicting Respiratory Function Based on Gait Analysis http://dx.doi.org/10.5772/63917 21

**Figure 22.** Different membership functions.

**Figure 21.** SVM classification result with *Dp*

20 Advanced Biosignal Processing and Diagnostic Methods

values come from ANFIS.

FVC FEV1, FVC, respectively. FEV1

FEV1

as, FEV1 ´ and *γ<sup>V</sup>* .

pulmonary spirometer. The ANFIS system comes from the toolbox of Matlab.

We utilize adaptive neural fuzzy inference system to help us predict the parameters from

Because we only collect about 60 cases so far, it is not enough for ANFIS to perform prediction. In order to increase the training samples, we adopt **Leave-one-out cross-validation** method. **Leave-one-out cross-validation** is used in analysing small datasets. It uses one sample as the validation set and the remaining as the training set. Repeat on this way for all samples. We

In ANFIS, it is important to choose a correct membership function. In addition, we also need to choose the input sections. In the experiment, we use six different membership functions including trapmf, gbellmf, gaussmf, gauss2mf, pimf and dsigmf. **Figure 22** shows the mem‐ bership functions we use in our prediction experiment. The inputs of the experiment are *Vp*

and *γ<sup>V</sup>* . In our results, we show the correlation, normalized Mean Square Error (MSE*N*) and regression slope under different membership functions. The formula of MSE*<sup>N</sup>* is shown in Eq.

We try to predict three different parameters that come from the pulmonary spirometer: post

*FVC* , postFEV1 and postFVC. The 'post' name means the parameters after the 6-min brisk

walking test. In the following part, for the convenience, we call post FEV1

so we choose these two parameters as our predicting targets.

(20). The Target*<sup>i</sup>* are the measured values from pulmonary spirometer and the Predict*<sup>i</sup>*

**6.2. Prediction with adaptive neural fuzzy inference systems**

can use this method to solve the insufficient data problem.


**Table 7.** The results of predicting FEV1 FVC .

´

are the

(20)

FVC , postFEV1, postFVC

FVC and FEV1 are used to access the respiratory function,

FEV1-FVC is an index which is used to access the severity of airway obstruction. The lower value means that the airway obstructs severely. **Table 7** shows our prediction results and we use [2 2] as the input sections. The *Vp* ´ and *<sup>γ</sup><sup>V</sup>* are the experiment inputs. **Figure 23** shows the predicting results and regression slope under different membership functions.

FEV1 is also a parameter to access the respiratory function. The higher FEV1 value means that the subjects have better respiratory function. Consequently, we also predict the FEV1 value. **Table 8** shows the prediction results and we use [3 2] as using 3 nodes for the first input and 2 nodes for the second input in the ANFIS system. The features *Vp* ´ and *<sup>γ</sup>V* are the experiment inputs. **Figure 24** shows the predicting results and regression slope in different membership functions.

**Figure 23.** The results of predicting FEV1 FVC with different membership functions: (a) Trap MF, (b) Gbell MF, (c) Gauss MF, (d) Gauss2 MF, (e) Pi MF and (f) Dsig MF.


**Table 8.** The result of predicting FEV1.

The correlation and regression slope of predicting FEV1 FVC do not perform well under all mem‐ bership functions. However, the correlation and regression slope of predicting FEV1 per‐ form well under trapmf. From the two results above, we infer that our system is good at predicting the respiratory parameter from pulmonary spirometer (FEV1). However, it is hard to predict the computed value ( *FEV* <sup>1</sup> FVC ). Consequently, we also predict the FVC value. **Table 9** shows the prediction results and we use [3 2] as using 3 nodes for the first input and 2 nodes for the second input in the ANFIS system. *Vp* ´ and *<sup>γ</sup><sup>V</sup>* are the experiment inputs.

**Figure 25** shows the predicting results and regression slope in different membership func‐ tions.

**Figure 24.** The results of predicting FEV1 with different membership functions: (a) Trap MF, (b) Gbell MF, (c) Gauss MF, (d) Gauss2 MF, (e) Pi MF and (f) Dsig MF.


**Table 9.** The result of predicting FVC.

**Figure 23.** The results of predicting

**Table 8.** The result of predicting FEV1.

MF, (d) Gauss2 MF, (e) Pi MF and (f) Dsig MF.

22 Advanced Biosignal Processing and Diagnostic Methods

FEV1

**Trapmf 0.694 0.140 0.746** Gbellmf 0.668 0.186 0.827 Gaussmf 0.640 0.193 0.774 Gauss2mf 0.386 1.968 1.264 Pimf 0.560 0.365 0.880 Dsigmf 0.352 2.692 1.321

The correlation and regression slope of predicting FEV1

2 nodes for the second input in the ANFIS system. *Vp*

hard to predict the computed value ( *FEV* <sup>1</sup>

**Correlation MSE***<sup>N</sup>* **Regression slope**

bership functions. However, the correlation and regression slope of predicting FEV1 per‐ form well under trapmf. From the two results above, we infer that our system is good at predicting the respiratory parameter from pulmonary spirometer (FEV1). However, it is

**Table 9** shows the prediction results and we use [3 2] as using 3 nodes for the first input and

FVC with different membership functions: (a) Trap MF, (b) Gbell MF, (c) Gauss

FVC do not perform well under all mem‐

´ and *<sup>γ</sup><sup>V</sup>* are the experiment inputs.

FVC ). Consequently, we also predict the FVC value.

**Table 10** shows the best results of predicting FEV1 FVC , FEV1 and FVC. The correlations of FEV1 and FVC are all good and close and the regression slope of FEV1 is better than FVC. However, the correlation and regression slope of FEV1 FVC do not perform well. The MSE*N* of FEV1 FVC is smaller than other two parameters. Our system is good at predicting the parameters of pulmonary spirometer but do not perform well in predicting the ratio. Nevertheless, the high correlations of FEV1 and FVC verify that there is a correlation between pulmonary spirometer and our gait analysis system.

**Figure 25.** The results of predicting FVC with different membership functions: (a) Trap MF, (b) Gbell MF, (c) Gauss MF, (d) Gauss2 MF, (e) Pi MF and (f) Dsig MF.


**Table 10.** The best result of predicting FEV1 FVC , FEV1 and FVC.

#### **6.3. Cooperating with radar system**

In this section, we combine the features of radar system with our features *Vp* ´ and *<sup>γ</sup><sup>V</sup>* . The radar system is a tool using impulse to record the moving of the subjects' chest and analyse the features of respiration. It uses the *Δ*Amp and the *Δβ*ratio to analyse respiration. These two features are listed in Eq. (21). The Amp is the respiratory intensity of the subject. The *β*1 and *β*<sup>2</sup> are the inspiratory speed and expiratory speed, respectively. The names 'post' and 'pre' mean the parameters after a 6-min brisk walking and before a 6-min brisk walking, respectively. We use these two features with *Vp* ´ and *<sup>γ</sup>V* to perform SVM classification and ANFIS prediction experiments.

In the SVM experiment, there are still 32 subjects in the Bad group and 28 subjects in the Good group. Because we use (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp,*Δβ*ratio) as the SVM inputs, we cannot draw a twodimensional (2D) figure. The accuracy with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio) is **81.6**% and it is higher than the accuracy with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ) (75%).

$$\begin{cases} \Delta Amp = \frac{postAmp - preAmp}{preAmp} \ast 100 \\ \Delta \beta ratio = \frac{post\beta ratio - pre\beta ratio}{pre\beta ratio} \ast 100 \\ \beta ratio = \beta\_1 \wedge \beta\_2 \end{cases} \tag{21}$$


**Table 11.** The best results of predicting FEV1 FVC , FEV1 and FVC.

î

of FEV1 and FVC verify that there is a correlation between pulmonary spirometer and our gait

**Figure 25.** The results of predicting FVC with different membership functions: (a) Trap MF, (b) Gbell MF, (c) Gauss

0.251 **0.040** 0.167

system is a tool using impulse to record the moving of the subjects' chest and analyse the features of respiration. It uses the *Δ*Amp and the *Δβ*ratio to analyse respiration. These two features are listed in Eq. (21). The Amp is the respiratory intensity of the subject. The *β*1 and *β*<sup>2</sup> are the inspiratory speed and expiratory speed, respectively. The names 'post' and 'pre' mean the parameters after a 6-min brisk walking and before a 6-min brisk walking, respectively. We

´ and *<sup>γ</sup>V* to perform SVM classification and ANFIS prediction

´ and *<sup>γ</sup><sup>V</sup>* . The radar

**Predicting target Correlation MSE***<sup>N</sup>* **Regression slope**

FVC , FEV1 and FVC.

In this section, we combine the features of radar system with our features *Vp*

FEV1 **0.694** 0.140 **0.746** FVC **0.678** 0.076 0.646

FEV1

analysis system.

24 Advanced Biosignal Processing and Diagnostic Methods

MF, (d) Gauss2 MF, (e) Pi MF and (f) Dsig MF.

**Table 10.** The best result of predicting

use these two features with *Vp*

experiments.

**6.3. Cooperating with radar system**

FEV1 FVC

> In the ANFIS experiment, we also predict FEV1 FVC and FEV1 parameters and use (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio) as the ANFIS experiment inputs. The input sections we used is [3 3 3 2] that means the number of nodes used in the four inputs are 3, 3, 3, and 2 respectively. **Table 11** shows the results of predicting FEV1 FVC , FEV1 and FVC with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ), (*Δ*Amp, *Δβ*ratio) and (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio). We only list the best results among the six different membership functions. **Figure 26** shows the best results of predicting FEV1 FVC , FEV1 and FVC with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ), (*Δ*Amp, *Δβ*ratio) and (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio). In predicting FEV1 FVC , the correlation and regres‐ sion slope improve strongly though the MSE*N* value increases slightly by using (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio). In predicting FEV1 and FVC, it does not improve the effects on

correlation and regression slope by using (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio). Therefore, the features of radar system cannot improve the results of predicting FEV1 and FVC.

Radar system improves our analysis results on both SVM classification and predicting the parameter *FEV1/FVC*. With radar system's help, there is a higher correlation and accuracy between the combined system and the pulmonary spirometer.

**Figure 26.** (a) Predicting FEV1 FVC with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ); (b)Predicting FEV1 FVC with (*Δ*Amp, *Δβ*ratio); (c) Predicting FEV1 FVC with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio); (d) Predicting FEV1 with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ); (e) Predicting FEV1 with (*Δ*Amp, *Δβ*ratio); (f) Predicting FEV1 with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio); (g) Predicting FVC with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* ); (h) Predicting FVC with (*Δ*Amp, *Δβ*ratio); (i) Predicting FVC with (*Vp* ´ , *<sup>γ</sup><sup>V</sup>* , *<sup>Δ</sup>*Amp, *Δβ*ratio).

#### **7. Conclusion**

We propose a vision sensor-based gait analysis method without wearing any sensor on human body. In our approach, the proposed gait features analyse the subjects' respiratory function. We also perform a clinical experiment on COAD patients and normal people with our vision sensor-based gait analysis method. With the extracted features, *Vp* ´ and *<sup>γ</sup><sup>V</sup>* , the classification result is close to the classification by the parameters of pulmonary spirometer. The SVM accuracy is **75%**. In ANFIS experiment, the correlations of ANFIS prediction on FEV1 and FVC achieve **0.694** and **0.678**.

In addition, by combining the features of radar system (ΔAmp and Δ*β*ratio) with our features (*Vp* ´ and *<sup>γ</sup><sup>V</sup>* ), the SVM accuracy and predicting on ratio ( FEV1 FVC ) both improve strongly. The SVM accuracy goes to **81** from 75% and the correlation of ANFIS on predicting FEV1 FVC goes to **0.428** from 0.25. From the experiment above, we verify that there exists a correlation between the pulmonary spirometer and gait analysis system.
