“The aim of medicine is to prevent disease and prolong life; the ideal of medicine is to eliminate the need of a physician.” - William J. Mayo, American Physician and surgeon and one of the 7 founders of the Mayo Clinic.
The cost of healthcare has been on the rise in this day and age, and your health has been the most valuable asset you own. Eating healthy and having a good sleep schedule is effective for preventing illness, but as people get busier, it has become more difficult to maintain your health. That is why it is essential for people to take dietary supplements. But how do people know what the best supplements are for them? They can ask a doctor, but the doctor might miss some information and excluding procedures or tests, you’ll have to pay an average of $80-$170 across the U.S for a doctor's visit.
That is why we are introducing Vital, an AI powered recommender for dietary supplements.
Vital uses users' personal physical information including their age, gender, allergies, and a text input where they write how they specifically want to improve their health. After entering those information, the users' information will be used to generate a list of the top recommended dietary supplement to take. This will help users save time, money, and confusion from spending time searching for which supplements to buy. The user will be able to enter new information to get supplement recommendations for different reasons and conditions.
We built Vital using the MERN stack, Redux, and sentiment analysis models for our ML. After the user enters their personal information, those data are used to generate recommendations. To give examples, if there are users filling out the form for a child under 6 years old who's allergic to peanuts, our model will avoid any supplements that are pills and contain peanuts. For the free response, if the use inputs "I want to have better skin", then our model will assess that sentence and recommend supplements that are useful for the skin. We trained our model by using an open source dietary supplement dataset.
One of the biggest challenges we faced was connecting the output of python files to our web application. We knew using flask would work best, but none of us had experience using it. We decided to use a node and express server to fetch the data generated by our python file. This required a lot researching and brute-forcing our way to success, which took a great amount of time. But we were able to overcome this challenge by each doing our research for each of our own parts and pair programming to make sure we implemented everything properly and efficiently.
We learned that it is possible to fetch data from a python file using a node and express backend server. Although flask may be more appropriate to use in that situation, we were able to leverage the skills and knowledge we already have about node and express to create our project with a tech stack we are already familiar with.
Vital will be further refined by training our ML model with better data and adding more features to the web app such as using more personal information to increase the accuracy of the recommendations.
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent