Nutrient Profile Model Assessment

For assessing individual nutrient scores of a product within the Nutrient Profile Model we have to complete the following steps:

Calculate compound A and C scores
Calculate a total Nutrient Profile Model score using A and C scores taking into account Fibre and fruit, vegetables and nut scores
Determine whether the product passes or fails with thresholds depending on whether the product is a food or drink

The nutrientprofiler package includes a high level wrapper function NPMAssess for handling this row-by-row on a data.frame with columns generated from NPMScore. Below we will explore the component functions of NPMAssess to understand how this function operates.

library(nutrientprofiler)

A and C score calculators

The first component of NPMAssess is the step to calculate the A and C scores. This involves two functions: A_scorer and C_scorer that accept a number of arguments corresponding to the appropriate nutrient score and sum them.

The A_scorer takes 4 arguments: energy_score, sugar_score, salt_score and fat_score. These are scores calculated in the previous NPMScore step and are summed in the A_scorer function to return a compound A score. The C_scorer takes 3 arguments: fvn_score (fruit, vegetable and nut score), protein_score and fibre_score and calculates the sum of these to return a C score.

A_scorer(energy_score = 3, sugar_score = 5, salt_score = 1, fat_score = 2)
#> [1] 11

C_scorer(fvn_score = 2, protein_score = 3, fibre_score = 1)
#> [1] 6

Nutrient Profile Model Score

After computing the A and C scores we’re able to calculate the overall Nutrient Profile Model Score of the product. The method for computing the score is conditional on the size of the A and C score and can in some circumstances take into account the fibre score and the fruit, vegetables and nut score. Therefore the NPM_total function takes 4 arguments: a_score, c_score, fvn_score, fibre_score.

NPM_total(7, 2, 1, 1)

# example using keyword arguments
NPM_total(a_score = 10, c_score = 4, fvn_score = 1, fibre_score = 2)

This function works by subtracting the C score from the A score. However, if the A score is greater than or equal to 11 and the fruit, vegetable and nut score is less than 5, the function performs a different calculation and substracts the sum of the fibre score and the fruit, vegetables and nut score from the A score. This follows the logic defined in the Technical Guidance for Nutrient Profile modelling.

After determining an overall score for the product within the Nutrient Profile Model the final step is to assess this score to determine if the product passes or fails. This is a simple test of whether the score is above or below a specific threshold where the threshold differs if the product is a food or drink.

If the product is a food an NPM score of 4 or more leads to a failure, whilst for a drink a product fails the NPM is it scores 1 or more.

NPM_assess works by accepting two arguments: NPM_score and type. type determines if the product is a food or drink by accepting either food or drink whilst NPM_score should be a numeric value calculated from the prior NPM_total step. The function returns a character vector with a single value of either "PASS" or "FAIL".

# pass
NPM_assess(3, "food")
#> [1] "PASS"

# fail
NPM_assess(6, type = "drink")
#> [1] "FAIL"