Nutrient profilings like Nutriscore and others in the world of AI: lack of efficacy and reductionism undermine those rudimentary tools

https://www.cambridge.org/core/journals/proceedings-of-the-nutrition-society/article/uses-of-nutrient-profiling-to-address-public-health-needs-from-regulation-to-reformulation/C72B273DF1D2DF151BF56BB656FF7BD3/core-reader


https://www.cambridge.org/core/journals/proceedings-of-the-nutrition-society/article/nutrient-profiling-for-product-reformulation-public-health-impact-and-benefits-for-the-consumer/CDCD9DC269CA4292DA20E5B795F37A01

https://www.sciencedirect.com/science/article/pii/S0899900716301587?via%3Dihub

https://www.ncbi.nlm.nih.gov/pubmed/28619781


Nutrient profiles are indeed the lists of ingredients for products and the nutritional labels. Nutritional scores are advice and so they are only the translation of what authors want to be guidance for their customers. For instance, if it is that goal:


"Their goal is to identify nutrient-rich foods, generally defined as those that contain more nutrients than calories and are low in fat, sugar and salt." 

The game is lost as obesity does not decrease if you eat low fat, low sugar and low salt. If you have any evidence in the last 55 years we will discuss them.
Indeed those scores are the tools of observational epidemiology which is a delusional pseudoscience when we deal with complex factors like diet.