^{1}

^{2}

^{3}

^{4}

^{1}

^{2}

^{3}

^{4}

In view of the lack of objective data support for product evaluation methods in the industry, a triangular verification method was proposed; it considered nursing beds as the study object and combined subjective evaluation with eye movement and electroencephalogram. Because the triangular validation method is based on the numerical value between the indicators and the frequency of ranking, this method is worth investigating for analyzing experimental data more scientifically. This paper focuses on the further analysis of the experimental data, especially the use of interval estimation method. After analysis, we obtain that proposal 2 is the optimal solution. This method is more suitable for product evaluation which will collect large amount of experimental data to obtain more accurate results. For industrial product designers, the evaluation of products by users is very important. In the design stage, how to grasp the user’s evaluation of the product more accurately is a difficult problem. This paper takes nursing bed as the research object and studies the user participation design in order to make the product more acceptable to most people after it is launched.

Nursing beds are designed as original ordinary steel beds, mechanical transmission beds, electric beds, or multifunctional beds. With the development of computer technology, development of multifunctional nursing beds is increasing. The development of multifunctional nursing beds is a breakthrough in realizing comprehensive nursing and is also an innovation in patient healthcare function [

The evaluation of nursing beds also focuses on functionality. Boorman et al. [

In terms of a product evaluation system, the current evaluation method is mainly based on expert opinions combined with random sampling. Subjective factors are major contributors in this evaluation method; it is impossible to determine whether the obtained evaluation is a true evaluation. Therefore, in a previous article, a triangular verification method, by combining subjective evaluation, electroencephalogram (EEG) data, and eye movement data, was proposed for a more convincing evaluation method [

Based on Kansei engineering [

Experimental process of the triangular verification model.

In Figure

Four proposed nursing beds.

The experimental method is shown in Figure

Experimental data acquisition model.

Data were obtained from 20 participants with normal vision. All the data in this experiment were obtained according to relevant standards. The experimental process is as follows:

Participants washed their hair with shampoo and dried it

Participants watched and understood the experimental guidance and signed the statement

Researchers prepared experimental instruments

Researchers explained the experimental process to the participants

At the beginning of the experiment, participants looked at the first randomly occurring proposal of the medical nursing bed and scored by pressing a button from 1 to 5 (1—worst; 2—worse; 3—normal; 4—better; 5—best). When the participants press the button, the first rendering experiment ends and the second rendering experiment begins until all the experimental materials are completed. To ensure the effectiveness of the experiment, after the first round of grading, four proposals will be played randomly. The experiment was repeated 50 times

After the experiment was completed, the eye tracker and brain instrument stopped recording

The experimental procedure is shown in Figure

Experimental setup.

According to the experiment, we obtain the following data:

The expected value of 20 people’s subjective evaluation is

The expected value of 20 people’s eye movement data is

The expected value of 20 people’s EEG data is

Electrode and topographic map.

Based on the above analysis, we can obtain matrix

The final data obtained are as follows:

The correlation analysis of the data is carried out, and the results are shown in Tables

Correlation analysis data 1.

e1 | e2 | e3 | F4 | F7 | F8 | FZ | ||
---|---|---|---|---|---|---|---|---|

1 | 0.903 | 0.693 | -0.998 | 0.919 | 0.966 | 0.862 | 0.944 | |

e1 | 0.903 | 1 | 0.866 | -0.926 | 0.963 | 0.903 | 0.973 | 0.949 |

e2 | 0.693 | 0.866 | 1 | -0.731 | 0.921 | 0.822 | 0.954 | 0.890 |

e3 | -0.998 | -0.926 | -0.731 | 1 | -0.939 | -0.973 | -0.890 | -0.959 |

F4 | 0.919 | 0.963 | 0.921 | -0.939 | 1 | 0.97 | 0.988 | 0.996 |

F7 | 0.966 | 0.903 | 0.822 | -0.973 | 0.97 | 1 | 0.921 | 0.988 |

F8 | 0.862 | 0.973 | 0.954 | -0.89 | 0.988 | 0.921 | 1 | 0.970 |

FZ | 0.944 | 0.949 | 0.890 | -0.959 | 0.996 | 0.988 | 0.970 | 1 |

Correlation analysis data 2.

e1 | e2 | e3 | F4 | F7 | F8 | FZ | ||
---|---|---|---|---|---|---|---|---|

FC1 | 0.729 | 0.951 | 0.914 | -0.767 | 0.895 | 0.767 | 0.951 | 0.855 |

FC2 | 0.773 | 0.905 | 0.993 | -0.806 | 0.960 | 0.883 | 0.979 | 0.938 |

FC5 | 0.825 | 0.859 | 0.946 | -0.847 | 0.962 | 0.940 | 0.944 | 0.959 |

FC6 | 0.835 | 0.784 | 0.865 | -0.847 | 0.922 | 0.948 | 0.875 | 0.936 |

FT9 | 0.872 | 0.994 | 0.915 | -0.899 | 0.972 | 0.898 | 0.990 | 0.952 |

FT10 | 0.887 | 0.872 | 0.902 | -0.903 | 0.971 | 0.976 | 0.937 | 0.979 |

F3 | 0.838 | 0.951 | 0.973 | -0.867 | 0.985 | 0.917 | 0.996 | 0.966 |

Correlation analysis data 3.

FC1 | FC2 | FC5 | FC6 | FT9 | FT10 | F3 | |
---|---|---|---|---|---|---|---|

FC1 | 1 | 0.917 | 0.81 | 0.687 | 0.969 | 0.784 | 0.937 |

FC2 | 0.917 | 1 | 0.971 | 0.904 | 0.943 | 0.943 | 0.992 |

FC5 | 0.81 | 0.971 | 1 | 0.98 | 0.891 | 0.991 | 0.964 |

FC6 | 0.687 | 0.904 | 0.98 | 1 | 0.81 | 0.987 | 0.898 |

FT9 | 0.969 | 0.943 | 0.891 | 0.81 | 1 | 0.891 | 0.975 |

FT10 | 0.784 | 0.943 | 0.991 | 0.987 | 0.891 | 1 | 0.95 |

F3 | 0.937 | 0.992 | 0.964 | 0.898 | 0.975 | 0.95 | 1 |

It can be seen from the table that there is a certain correlation between subjective evaluation data, eye movement data, and EEG data. The first fixation time was negatively correlated with other indicators because the shorter the first fixation time, the more attention the participants paid, and vice versa.

Then, we need to analyze the results in terms of three different factors, namely, subjective evaluation, eye movement, and EEG data. First, proximity analysis is carried out.

To calculate the maximum

To calculate the distance from the maximum to minimum of each proposal.

To calculate relative closeness of evaluative value and maximum value for each program.

We use the relative closeness of evaluative value and maximum value for each proposal as the foundation of the final evaluation for the design proposal.

Through the above steps, we can obtain the results of the close degree analysis data as shown in Table

Data table for proximity analysis.

Index | C1 | C2 | C3 | C4 | Proposal sorting |
---|---|---|---|---|---|

0.84 | 1 | 0 | 0.55 | C2>C1>C4>C3 | |

e1 | 0.44 | 1 | 0 | 0.4 | C2>C1>C4>C3 |

e2 | 0.33 | 1 | 0.12 | 0 | C2>C1>C3>C4 |

e3 | 0.21 | 0 | 1 | 0.48 | C2>C1>C4>C3 |

F4 | 0.55 | 1 | 0 | 0.23 | C2>C1>C4>C3 |

F7 | 0.79 | 1 | 0 | 0.3 | C2>C1>C4>C3 |

F8 | 0.41 | 1 | 0 | 0.19 | C2>C1>C4>C3 |

FZ | 0.64 | 1 | 0 | 0.26 | C2>C1>C4>C3 |

FC1 | 0.16 | 1 | 0 | 0.24 | C2>C4>C1>C3 |

FC2 | 0.37 | 1 | 0.03 | 0 | C2>C1>C3>C4 |

FC5 | 0.6 | 1 | 0.05 | 0 | C2>C1>C3>C4 |

FC6 | 0.82 | 1 | 0.06 | 0 | C2>C1>C3>C4 |

FT9 | 0.39 | 1 | 0 | 0.31 | C2>C1>C4>C3 |

FT10 | 0.7 | 1 | 0 | 0.08 | C2>C1>C4>C3 |

F3 | 0.4 | 1 | 0 | 0.11 | C2>C1>C4>C3 |

Then, we use the frequency statistics method for the three factors and obtain the final ranking method according to the frequency of the four rankings. The specific calculation formula is as follows:

The frequency of proposal

Frequency table of subjective evaluation.

Subjective evaluation proposal ranking | First place | Second place | Third place | Fourth place | ||||
---|---|---|---|---|---|---|---|---|

C1 | 0 | 0.00 | 1 | 1.00 | 0 | 0.00 | 0 | 0.00 |

C2 | 1 | 1.00 | 0 | 0.00 | 0 | 0.00 | 0 | 0.00 |

C3 | 0 | 0.00 | 0 | 0.00 | 0 | 0.00 | 1 | 1.00 |

C4 | 0 | 0.00 | 0 | 0.00 | 1 | 1.00 | 0 | 0.00 |

Final ranking | C2 | C1 | C4 | C3 |

Frequency table of eye movement test evaluation.

Subjective evaluation proposal ranking | First place | Second place | Third place | Fourth place | ||||
---|---|---|---|---|---|---|---|---|

C1 | 0 | 0.00 | 3 | 1.00 | 0 | 0.00 | 0 | 0.00 |

C2 | 3 | 1.00 | 0 | 0.00 | 0 | 0.00 | 0 | 0.00 |

C3 | 0 | 0.00 | 0 | 0.00 | 1 | 0.33 | 2 | 0.67 |

C4 | 0 | 0.00 | 0 | 0.00 | 2 | 0.67 | 1 | 0.33 |

Final ranking | C2 | C1 | C4 | C3 |

Frequency table of EEG test evaluation.

Subjective evaluation proposal ranking | First place | Second place | Third place | Fourth place | ||||
---|---|---|---|---|---|---|---|---|

C1 | 0 | 0.00 | 10 | 0.91 | 1 | 0.09 | 0 | 0.00 |

C2 | 11 | 1.00 | 0 | 0.00 | 0 | 0.00 | 0 | 0.00 |

C3 | 0 | 0.00 | 0 | 0.00 | 3 | 0.27 | 8 | 0.73 |

C4 | 0 | 0.00 | 1 | 0.09 | 7 | 0.64 | 3 | 0.27 |

Final ranking | C2 | C1 | C4 | C3 |

From Tables

Because the original triangular validation method mainly relies on the size of the data value to arrange the data, which is not convincing to a certain extent, there is contingency; therefore, this study focuses on data processing, especially of the confidence interval validation method used in data analysis. We assume that the reader is familiar with the basic notions of statistical theory.

According to central limit theorem [

The specific data of proposal 1 of the F3 indicator are as follows:

The box diagram of the F3 index is shown in Figure

Box diagram of the F3 index.

The final data and sample size

F3 index data sheet.

Participant | F3-1 | F3-2 | F3-3 | F3-4 |
---|---|---|---|---|

1 | 2.842 | 3.203 | 4.900 | 5.187 |

2 | 3.406 | 2.500 | 1.314 | 1.233 |

3 | 0.996 | 4.231 | 0.898 | 3.546 |

4 | 0.806 | 0.514 | 0.095 | 0.174 |

5 | 3.580 | 1.805 | 1.460 | 1.341 |

6 | 2.256 | 1.989 | 1.284 | 2.605 |

8 | 0.435 | 0.674 | 0.457 | 0.875 |

9 | 0.943 | 3.273 | 1.867 | 3.999 |

10 | 1.165 | 2.203 | 1.410 | 1.482 |

11 | 0.991 | 1.385 | 0.568 | 0.438 |

12 | 0.824 | 2.206 | 2.159 | 1.108 |

13 | 3.616 | 4.230 | 2.686 | 1.148 |

14 | 1.158 | 1.287 | 0.722 | 1.596 |

15 | 1.586 | 4.807 | 2.457 | 1.241 |

17 | 1.447 | 3.501 | 1.076 | 0.881 |

18 | 3.189 | 2.851 | 2.219 | 1.999 |

19 | 1.914 | 1.530 | 0.904 | 1.097 |

20 | 1.321 | 3.301 | 1.733 | 2.500 |

18 | 18 | 18 | 18 | |

1.804 | 2.527 | 1.567 | 1.803 | |

1.067 | 1.237 | 1.097 | 1.309 | |

1.138 | 1.530 | 1.204 | 1.713 |

According to Table

The

Distribution

Thus, we get a confidence interval of

According to the data table of

that is,

Because the confidence interval contains 1, we can assume that

Variance confidence interval data in the F3 index.

Lower bound | 0.393 | 0.500 | 0.351 | 0.673 | 0.473 | 0.372 |

Upper bound | 1.405 | 1.786 | 1.255 | 2.402 | 1.688 | 1.328 |

Based on the above analysis, we may consider the variance of each set of data to be equal and assume that the population variance of each data is equal. Next, we will verify the confidence interval of the population mean difference

or it can be written as

A confidence level of

Although

In this formula,

According to Table

Based on the above formulas, we can obtain the confidence intervals as

that is,

Because the lower bound of the confidence interval is greater than zero,

Confidence interval data of mean in the F3 index.

Lower bound | -0.004 | -0.467 | -0.737 | 0.229 | -0.507 |

Upper bound | 1.450 | 0.942 | 0.740 | 1.692 | 0.979 |

Similarly, in proposal 3, proposal 1, and proposal 4, the lower bound of confidence interval is lower than zero; therefore, we can conclude that according to the F3 index in this analysis method, there is no significant difference between proposal 3, proposal 1, and proposal 4.

The same analysis method was adopted to verify the other 14 indexes; the nine indicators cannot draw a clear conclusion. Specific conclusions are shown in Table

Confidence interval data of mean in indexes.

Lower bound | 0.071 | 0.278 | 0.584 | 0.240 |

Upper bound | 1.671 | 1.828 | 2.033 | 1.878 |

Lower bound | 0.031 | 0.276 | 0.197 | |

Upper bound | 1.945 | 2.181 | 1.988 |

According to the data obtained, we can see that proposal 2 is better than the others, that is, proposal 2 is the best, which is consistent with the previous conclusion.

In this study, we improved the way of data comparison in the triangular validation method. We apply the confidence interval in statistics to the analysis model and improve the problem of previous data comparison which is being too simple. Because less amount of data is collected in this experiment, a definite result cannot be acquired from a single indicator. However, according to statistical theory, when the number of data sample is large, the model constructed will draw a clear conclusion. This study shows that mathematical statistics can be well used in product evaluation and that triangular evaluation is deepened to make the evaluation model more convincing and applicable. With the development of intelligent wearable equipment, data acquisition will become more convenient in the future. Therefore, this evaluation model should have an extensive application and research value. We hope that more mathematical and statistical knowledge will be used for product evaluation to promote the development of industrial evaluation systems. Next, we will analyze the product evaluation in industrial design and guide the product design process through the analysis of a large number of data.

No data were used to support this study.

The authors declare that they have no conflicts of interest.

The work presented in this paper is supported by Shanghai Multidirection Die Forging Engineering Technology Research Center (No. 20DZ2253200), National Key R&D Program—Research on Modernization of Traditional Chinese Medicine (2018YFC1707802), and science and technology support project in biomedical field of “Science and Technology Innovation Action Plan” of Shanghai in 2019 (No. 19441914900).

^{2}- and F-distributions and their applications