Hello, this is @doctorbme. Today, we will see what type of information on user's condition and activity can be obtained in terms of designing an AI for meal planning and how it can be applied. Whereas the previous article was about utilizing the subjective information that reflects user/expert opinions, this article presents various types of information that can be identified within the meal planning AI service itself. ************************************************************************************ First, let's briefly observe what types of factors on the meal planning AI service. The factors below may not be perfect, but it is a general idea of which factors should be considered. ![](https://cdn.steemitimages.com/700x0/https://cdn.steemitimages.com/DQmPpHEGbvuf38WWDr79Fcf2o6YuorAPcRgXf7UcBXnbR6f/image.png) Figure 1. Factors of meal planning AI service 1) Database on menu, food, and ingredients: This is the basic information for making food. It may include ingredient information such as calories, nutrients, etc., food that can be made when ingredients are combined in the right order, and relevant recipes. 2) Algorithm: It shows the algorithm for finding the optimum menu per user by considering the limiting conditions that reflects the status of each user based on the given database. 3) Information processing and adaptation: Although differences can occur depending on which algorithm is implemented, it shows new recipes or menus can be re-saved in the database, or the existing information found in the database can be reprocessed or retrieved according to the situation. In addition, the information identified based on various information sources other than the information directly entered by the user can be processed and adapted according to the database. 4) Direct input of information: It reflects the subjective opinions of the user/expert. 5) Information acquisition: It reflects various objective information related to user's smartphone sensor, blood glucose monitor, blood pressure gauge, etc. 6) Result configuration and output: It outputs configured diet. **************************************************************************************** For example, let's suppose that a diabetes patient uses such meal planning AI. Then, Additional information is required beyond the information obtrained grom the general user. For ingredients and food, the influence of food on blood sugar should be considered based on the glycemic index (GI). If the model is expanded a little further, compartment modeling can be used to configure the diet with limiting conditions of maximum and minimum blood sugar by predicting and observing the changes in blood sugar, which can be done by creating a self-model of how the blood sugar and insulin levels of the user are changed when food with certain GI and various nutritional content is consumed. ![](https://cdn.steemitimages.com/700x0/https://cdn.steemitimages.com/DQmX6npzwoktMAj9j7fmHduceyJgdorSQKZtLA3QD6MqDut/image.png) ************************************************************************* Subjective opinions that the user can set would be how much I want to gain/lose weight, whether to prefer/not prefer food containing certain ingredients, whether to accept new recipes, etc. Then, what is some objective information that can be obtained from the user? Information that can be measured via the user may include not only weight, body mass index (BMI), and regularly checked blood sugar, but also basal metabolism that reflects muscle mass, etc., measurement of motion quantity via user's movement linked to smartphone's GPS and giro sensor, type and configuration of food consumed if the user is keeping a record of consumed food, each users genetic information, metabolic process which microorganisms and food react to, etc. Especially for the latter, ![](https://cdn.steemitimages.com/700x0/https://cdn.steemitimages.com/DQmbRgenvuAjJ4eMrsusmhAbatenrWVVN7wv3A3Q3KpcYoV/image.png) As such meal planning includes the management and monitoring of user's health, it shows the potential application in overall health management including both exercise and meal planning. ********************************************************************************************************** Currently, various services related to AI are being developed. FitGenie Provides diets by reflecting various factors such as hunger and exhaustion, and improves via feedback Nutrino In addition to personal status and information, attempts to converge various literature on diets and knowledge based on natural language processing. NOT company Attempts to find combinations of ingredients that can replace existing meat-based diet with plant-based diet while similarly maintaining food, flavor, and nutrition. Nutritas Aims to find various biochemical bioactive agents contained in food ingredients related to health improvement. SmartPlate Attempts to present a better diet by recognizing photos of food, calculating calories and nutrition, and analyzing diets. A variety of other AI-based food-tech services are also on the rise. The main characteristics of such services are that they are aimed toward 1) managing users' health or 2) discovering new food ingredient/food combinations. With this, I would like to sum up the introduction of AI on meal planning. In fact, with close observation, there are many things to be considered and various services that need to be benchmarked. I just hope the above information can help you create the big picture on how to design an AI service for meal planning. Thank you. ******************************************************************************************************* Reference [1] Rodrigo Zenun Franco, Rosalind Fallaize, Julie A Lovegrove, and Faustina Hwang, Popular Nutrition-Related Mobile Apps: A Feature Assessment, JMIR Mhealth Uhealth. 2016 Jul-Sep; 4(3): e85. [2] Sonnenburg JL, Bäckhed F, Diet-microbiota interactions as moderators of human metabolism, Nature. 2016 Jul 7;535(7610):56-64. doi: 10.1038/nature18846. [3] J. Bulka, A. Izworski, J. Koleszynska, J. Lis, I. Wochlik, Automatic meal planning using artificial intelligence algorithms in computer aided diabetes therapy, IEEE ICARA 2009, https://doi.org/10.1109/ICARA.2000.4803989 *************************************************************************************************** All the pictures used in this article are self-produced ********************************************************************************************** [![](https://cdn.steemitimages.com/700x0/https://cdn.steemitimages.com/DQmb3ZomKS1sGpntm5Tdsx4Sr1vcDt9vSk47rp7y7Nd73ku/image.png)](https://mediteam.us/) The first STEEM medical magazine written by professionals