calibrateSlope: Calibrate a loglinear Slope

In uvic-cisur/intermahpr: The International Model of Alcohol Harms and Policies

Description

Given a list that contains the true count among a given cohort of a wholly alocohol-attributable condition, the number of drinkers in that cohort, the conditions's IM code, and the gamma function detailing consumption among the cohort, this function calibrates a slope for a loglinear function that estimates the conditional probablity of developing the given condition. The function should be interpreted as conditional probability mass. I.e., the integral of the function over the relevant range is the probability that a drinker will be afflicted by an event of the given condition over the period of time that the given Count variable was collected.

Uses a nonlinear optimizer (COBYLA) to find a loglinear slope for the function f(x) = max(1, exp(k(x-t))) that mimizes the difference between integral(N_GAMMA * (f-1), 0.03, UB) and yearly prevalence (count/drinkers).

The goal is to produce a continuous analogue to the relative risk curve for conditions that are wholly attributable to alcohol. The assumption is made that such a condition has a loglinear thresholded (i.e. f(x)=1 for x<t) conditional probablity function on the interval of concern (0.03 to UB grams of ethanol/day, averaged over 1yr).

This conditional probability is used to portion a 1.00 AAF_TOTAL among the drinking population.

Usage

calibrateSlope(target, mass, lb, ub)

Arguments

target	Observed incidence to calibrate against
mass	population exposure mass function to calibrate against
lb	lower bound of consumption at which condition occurs
ub	upper bound of consumption

Value

slope of loglinear conditional probability mass function for risk as a result of exposure

uvic-cisur/intermahpr documentation built on June 17, 2020, 1:30 a.m.