Abstract
This study was conducted to get a resonable set of budget allocation to public health programs. Matrix Delphi technique was used to obtain the logic of study results and eventually to form a human model which could predict opinion of professionals on budget allocation. Thirty-two professionals in academic and governmental area responded to Delphi survey. Questionnaire was developed using matrix formation, and the matrix was formed by 6 decision criteria on budget allocation and 26 public health programs. The decision criteria are as following: size of problem(morbidity), severity of problem, social equity, importance of prevention, technical feasibility and efficiency of programs. Severity of problem dropped out of the model because it had significant correlation with the size of problem. A total score of each program was obtained by weighting the relative importance of each criteria which also were given by survey respondents. These total scores indicate that the most important public health program is vaccination for infants and children in terms of budget allocation. Monitoring communicable diseases, mental health program, and anti-smoking program are the next. In addition, respondents were asked of the desirable budget size of each program. The result was rearranged by multiple regression model using the scores of each decision criteria. In this process, the current budget size of central government was provided to the respondents, and included in the model. h set of desirable budgets modified using tile model was obtained. Considering the current size of budget, tile results of the model is very different from that of the total score. Managing dementia is ranked the first. Health promotion program for the elderly, rehabilitation of the disabled and monitoring communicable diseases are the next. The need to increase the budget of vaccination for the infants and children was not found as so high. The matrix structure in Delphi survey gave us the precise basis to make optimal decision, and made it possible to develop an opinion predicting model. However the plentifulness and diversity of professional opinions were not fully obtained due to the limited number of decision criteria.