Forecasting median and mode dates of prevalence of Japanese encephalitis patients by electronic computer (epidemiological studies on Japanese encephalitis, 31)

DOI 機関リポジトリ

この論文をさがす

抄録

<p>For the purpose of forecasting the prevalence ofJapanese encephalitis in Japan, we tried to find out the correlation of factors between median and mode dates of epidemic time curve of prevalence on one hand, and average atmospheric temperatures of prefectures in June and July (T6,7 in short) (X¹), the time when HI reaction of swine became positive to the degree of 50 per cent (D. pos. swine in short) (X²), the latitude (x³) and longitude (x4) in respective prefectures (in 1965 and 1967). On the other we also estimated the median and mode dates of this epidemic curve of the prevalence in 1968 and 1969, from the regression equation of one variable and multiple regression equation from the above factors using an electronic computer. The usefulness of adding factors concerned with mosquitoes to the above four factors is proven by the accuracy of estimation. And the following results were obtained. 1) Phenomenally speaking, the prevalence of Japanese encephalitis follows the principle of "advancing of prevalence towards the north and east" and essentially speaking, it depends upon high atmospheric temperature and the outbreak of many hazardous mosquitoes by the high atmospheric temperature. 2) To estimate median date (y) and mode rate (z) of the epidemic time curve of the prevalence, we can use the next equations; The regression equations to estimate y and z from T 6,7(X) are as follows. y = - 3. 75X¹ + 144.47 σ = 12.4.·. [1] z = - 3. 80X¹ + 157 .26 σ = 14.9.. · [1]' The regression equation from D. pos. swine (X²) are as follows. y = 0. 68X² + 31. 82 σ = 9.2· .. [2] z=0. 76X² +40. 71 σ= 12.0 .. · [2]' The multiple regression equation from T6 ,7 and D. pos. swme are as follows. y = -1. 07X¹ +0 .62x² +59. 37 σ= 9.7 ... [3] z= -0. 79x¹ +0. 71x² +61.02 σ= 12.0· .. [3]' The multiple regression equations from T 6•7, D. pos. swine, latitude and longitude are as follows Y= -1.01x¹ +0.58x² -0.26x³+0 .37x4 + 18.50 σ= 9.8･･･ [4] z = -0. 32x¹ +0. 52x² +2 .05x³ +0 .54x4 -87. 81 σ= 11.8 [4]' 3) We Obtained the estimated value of median date in 17 prefectures in Kyushu, Chugoku, Shikoku, Kinki and Kanto provinces in 1968 and in 13 prefectures in 1969 from [l] or [2] or [3] or [4] equation. Nine prefectures out of 17 by [l], 12 prefectures by [2], 13 by [3J and [4] in 1968. [4] could be estimated with about 10 days error or less. And in 1969, 9 out of 13 by [3] and 7 out of 13 by [4] could be accurately esti· mated. The estimation by the multiple regression equation using many factors is most useful for the calculation. 4) The time when the number of patients increases at maximum can be pointed out by the lower limit of prediction region obtained from data in each prefecture. And the lower limit was the estimated median value minus about 20 days by [1] and about 16 days by [2] or [3] or [4] under the next condition; α = 0. 1, N= 75. 5) The mode dates in 17 prefectures out of 19 were estimated by [1]', [2]', [3]' and [4]'. 12 prefectures out of 17 by [1]', 7 by [2]', 10 by [3]' and 13 by [4]' could be estimated with about 12 days error or less in 1968 and 9 out of 13 was correctly estimated by [3]' and [4]' in 1969. The estimation by the regression line of one factor was s~mewhat different from each other, but when multiple regression line of four factors was used the estimation became more correct. Judging from these results, it is adequate to use the multiple regression equation of [4] and [4]' when we want to forecast the median date or mode date ofJapanese encephalitis time cure. 6) In the case of adding two factors concerned with mosquitoes to T6,7 (X¹), D. pos. swine (x²), latitude (x³), longitude (x4), multiple regression equations become as follows. y= -1.46x¹+0.14X²+0.068x5+89.03 σ= 6.9.. ·[5] z= -3. 29x¹+0 .13x²-0. 010x5+ 143.63 σ= 18.6··· [5]' y=-4.20x¹+0.35x²+0.29x6 + 53.70 σ= 4.2 .. ·[6] z=-2.56x¹-0.0lx²-0.02x6 +128.96 σ=11.4 [6]' y= 4.76x¹+0.41x²+0.13x5+0.22x6-72.78 σ= 4.5 [7] z = - 2. l0x¹ + 0. 05x²+ 0. 11 x5 - 0. 08x6+ 113.4 σ= 10. 7.. · [7]' where x5 is the time when the number of mosquitoes (C. T. collected by light trap reached the maximum and X6 is the time when hazardous mosq uitoes were dected. In the case of median date, 5 prefectures out of 6 prefectures by [5], 2 out of 6 by [6] and 2 out of 5 by [7], and in the case of mode date, 5 out of 6 by [5]', 4 out of 5 by [6]' and 4 out of 5 by [7]' could be accurately estimated in 1969.</p>

収録刊行物

詳細情報 詳細情報について

問題の指摘

ページトップへ