【正文】 it can also measure the effective amount of information of the data. If there are m criterions and n objects which need to be evaluated, the entropy of the thi criterion is defined asjH: 1 lnnj ij ijjH k f f?? ? ?? ( 1, 2, )jm??， , (6) in this formula: nkrrfnj ijijij ln1,1?? ?? ? The choice of metrics For the first principal ponent, the correlation between life expectancy, infant mortality rate and neonatal mortality rate is significant, while the life expectancy has the biggest Information Entropy, so we chosen life expectancy. For the second principal ponent, the correlation coefficient of mortality rate of typical epidemic sufferer and typical epidemic morbidity is , while the latter has the bigger Information Entropy, so the typical epidemic morbidity is chosen. Similarly, for the third principal ponent, the correlation coefficient of density of physicians, nurses, etc and hospital beds per 1000 population is , but the latter has the bigger Information Entropy, so the hospital beds per 1000 population is chosen. Finally, for the fifth principal ponent, the correlation matrix is shown as table 7 displayed below. It’s observed that the correlation between total expenditure on health as %of GDP and per capita total expenditure on health is significant, we need to choose only o ne of them. At the same time, the correlation government expenditure on health as %of total expenditure on health and private expenditure on health as %of total expenditure on health is linear, since the Information Entropies of total expenditure on health as %of GDP and government expenditure on health as %of total expenditure on health are bigger, we chosen both of them. The conclusion of chosen important metrics To make the result more clear, we listed them in one table (refer to table 8) Table 8. Metrics chosen with the Components Component The most important metrics Table 7. Correlation Coefficient Matrix of Financial Policy Metrics Coefficient Total Government Private Per capita Total 1 Government 1 1 Private 1 1 Per capita 1 第 13 頁 共 26 頁 General health status life expectancy The overall epidemic condition typical epidemic morbidity Resources of health systems hospital beds per 1000 population Financial Policy total expenditure on health as %of GDP Government expenditure on health as %of Total expenditure on health Historical changes of the system The change reflected on the share of GDP along with the improvement of people’s life, more and more attention are paid to the health care systems, which can be reflected by the fact that the pace spending money on health care systems bee faster and faster. Take the USA for example (shown in figure 2) Figure 2. the Change of share of GDP in the USA It can be shown that the share of GDP in the USA is increasing in the recent years, which can reach to 20% in 2021. The change of life expectancy since the long life expectancy is the aim of every health care system, it can reflect the change of a system deservedly. Take China for example, the life expectancy have increased years from 1990 to 2021, which means the life expectancy in China has reached to in 2021. what’s more, in other developed countries, take England, America and France for example, the average life expectancy of them increases years each year. All the changes mentioned above prove that the health care system in these countries is being better and better. change of hospital beds per 1000 population As we all know, the hospital beds can reflect the hardware of a certain system to a certain extend. The more hospital beds a person has, the better the system is. In fact , the number of。 on the other hand, it will be changed along with the development of economy in a potential system. What’s more, generally speaking, the more the cost is, the better effectiveness is, which means the change of effectiveness dependents on the cost’s change to a great extent. Thus the first step to make parisons between existing and potential systems is to divide the metrics into cost and effectiveness, and then use the cost to determine which system is better. Classing the chosen metrics The costs here mainly refer to the measures which can be change by people, such as how much can be acplished with currently available resources – people, buildings, equipment and knowledge – depends greatly on the past investment and train in that created those resources. Corresponding with the metrics we have chosen, the costs are health systems resource, health systems expenditure and inequities. Since them can be change by people and can affect the effectiveness of a systems directly, we use them to pare the current and potential systems. By all appearances, the rest metrics we have chosen should be the effectiveness. Conclude the metrics used to pare Based on the analysis above, the metrics used to make parisons between existing and potential systems are shown in the table 2 Table 2. Metrics Used to Make Comparisons Health systems resource human resource of Health systems as %of total population Per capita material resources of Health systems Health systems expenditure Total expenditure on health as %of GDP Government and Private expenditure on health as %of Total expenditure on health Per capita total expenditure on health Inequities Inequities in health Combining the chosen metrics Refer to the above analysis that the costeffectiveness analysis is essential for identifying the systems that will produce the most health gain from available resources, hence the sort we have classed can also be used in this problem. 第 5 頁 共 26 頁 For the metrics included in the cost, we can bine them to get a positive cost index, similarly, ano