Blood Glucose Simulator

This is a web application to track daily blood glucose levels.

Setup Instructions

To run, download or clone the repo. Then, issue the following commands:

bundle install
rails db:migrate
rails db:seed
rails server

To run the test suite, run the following command:

bundle exec rake test

Requirements

eating food increases blood sugar linearly for 2 hours, depending on the food's glycemic index
exercising decreases blood sugar linearly for 1 hour, depending on the exercise's index
blood sugar starts at 80
if no activity or food for a while, blood sugar normalizes to 80 linearly at a rate of 1/minute
for every minute the blood sugar is above 150, glycation increments by 1
inputs: food and exercise data (strings)
outputs: graph of blood sugar for the day and a graph of glycation for the day

Design Docs

Features

My minimum viable product allows users to

enter in food eaten or exercise done (will match to database entries if they exist or create new items if they don't exist)
look at graphs, utilizing the Gruff gem, of blood glucose levels or glycation levels per day

Design choices

I chose Ruby on Rails as a framework because it is very fast for creating MVPs. Initially, I was going to save food and exercise items to the DB as they were entered, while calculating and saving blood sugars to the DB as well, based on those items and all items up to that point starting from the beginning of the day. However, that would result in a lot of blood sugar items being saved to the DB as I was planning to save values for each minute. I was either going to do a cron or background job for the saving. But, I found out that that would take too much time to implement and cause heavy load on the DB. Forgoing that, I decided to try another approach. I saved food and exercise entries as events with the relevant glycemic or exercise indices. Then, I grabbed all events past midnight, calculated increments that each event would cause, and then calculate the blood sugar for all minutes of the day in memory.

Future considerations to scale current implementation

have a background job calculate the blood sugar levels
use a cache for the latest values so that visiting the graph page doesn't trigger recalculations every time
add indexes on search terms for faster lookup
use 3rd party API to estimate glycemic or exercise index of new items not in the DB
improve test coverage for algorithm/calculations
improve the user interface with an autocomplete for food/exercise items that already exist and a more interactive graphing library
use design patterns to abstract out common code (search, graph, and calculations) to helper files or modules