THE 'COLLIER' EXTRAPOLATION METHOD

OF FORECASTING AIDS/HIV, ETC., BY USE OF AN

INFINITE SELECTION OF QUASI-EXPONENTIAL, COMPUTERISED FRENCH CURVES.
SYNOPSIS The method of forecasting described in this article was devised by the author early in 1987 and has been used ever since by him in the forecasting of the growth of the AIDS epidemic in the UK. It has also been used for forecasting UK variant CJD. Forecasts of AIDS/HIV in the UK can be found on this website, both quarterly comprehensive forecasts since the end of 1999 and the 9^th Annual Report at the end of 1999.
(It is regretted that some of the numbers and sections of text of the illustrations have been degraded by the conversion from .PRE to .PRZ to .TIF and finally to .GIF)

HISTORY This extrapolation method is, in effect, a method of applying computerised French Curves to a group of the latest data reported by the Public Health Laboratory Service. The ‘best-fit’ curve is then extrapolated to provide projected values into the future. The simple method was described at an AIDS conference in 1987 that was held under the auspices of the Royal Statistical Society. The 21 quarterly forecasts from end-September 1991 to end-September 1996 showed excellent consistency with mean value for Slope of 0.76396922 (Std. Dev. 0.005) and - 60.309954 (Std. Dev. 0.4675) for constant. Figure 1 below shows how well the French Curve, selected by that mean doubling time, actually represents the actual reported growth of the cumulative numbers of AIDS cases in the UK.

( 1 )

NEXT HOME