What it does
Vita-Cam uses Artificial Intelligence to detect possible vitamin deficiencies by asking the user to take photos of four body parts: Eyes, Lips, Tongue and Nails. The AI then uses the analytical information to show suitable diets and supplements accordingly!
Vitamin deficiency underlines hundreds of health issues presented in our daily life, in which many health problems arise from the failure in acquiring the necessary spectrum of the vital minerals and nutrition. It may be difficult to keep track of our nutritional needs especially when individuals are not aware of the type of deficiency they are holding without medical consultation. Over 2 billion people worldwide suffer from vitamin deficiencies. More than 1.2 billion individuals are Zinc deficient in which half a million of them die each year. Similarly speaking, over 100,000 people die due to Anemia caused by Iron deficiency.
How it works
An Android platform suitable for acquiring, analyzing, detection and expanding was designed using the combination of two advanced mathematical models of Artificial Intelligence & Machine learning: Convolutional Neural Network (CNN) and Fuzzy Logic. After conducting an extensive medical research, a neural network was calibrated and trained to be able to extract certain features (symptoms) from imagery inputs - in terms of confidence levels - by performing pixel convolutional mathematical operations on hundreds of carefully selected medical conditions photos. The values of each estimation of this computer-vision based analysis are then fed into a Fuzzy Logic decision making algorithm with pre-set decision rules to specify the certainty of the detected vitamin deficiency(s) and display the final result of detection; followed by a list of nutritional sources known to be rich in the deficit vitamin and a page to allow companies to advertise their products.
First, A medical study was conducted to build a relationship between symptoms and vitamin deficiencies on selected spectrum of visually distinguished attributes. The four previously mentioned body parts were chosen as they show changes in texture, shape, color or appearance when an insufficiency in one or more of the essential vitamins is presented. A database containing collected photos showing these symptoms have been constructed to perform feature extraction neural training. The trained TensorFlow classifier was imported and implemented in an android environment using Android Studio. The application prompts the user to take four separate pictures of their tongue, lips, eyes and nails. The images are saved with full resolution in an internal directory and called separately to perform the neural feature analysis and extraction. A Fuzzy Inference system was built to consider confidence rates higher than 75% as high influence weights on Mamdani’s center of gravity. The Defuzzification result is processed to display details about the associated deficiencies to the user which is done by the integration of the Fuzzy Logic algorithm in a C++ format using MATLAB Coder, which allows the extraction of the Fuzzy Membership Functions and Inference to a compact C language library or Java.
How it is different
The difference is that there was no similar approach to this issue before, and it is safe to say that this is the first attempt of its kind. The power factor in this project is the low cost and portability as it is targeting everyday individuals and users with less to no technical nor medical experience needed. it is not meant to replace medical checkup - at least for now! -, but it is designed to provide a way for people to keep track of their body needs without spending hefty amount of money. Our philosophy is simple: "prevention is better than cure", so when everyone is able to self-asset her/his situation independently and self-correct their nutritional habits toward a healthy plan, the negative impact of the deficiency will gradually diminish. In a survey we have conducted, about 67% of people reported that they were not aware of their vitamin deficiencies, so how about we give them a simple tool to evaluate and take care of themselves better?
The second stage is to work on modifying the platform to allow medical experts and research centers to assist in improving the range of detection and accuracy of the application by adding visual data of their patients; allowing the platform to build more delicate image analysis capabilities that might surpass human diagnostic. Our main customers are dietitians and supplement companies, Vita-Cam will be the link between ordinary people and these firms. Profit would be either from adds and order commissions or structuring the app in a way to attract highest number of customers with limited charges with the goal to be sold to a larger system.
1st place UAE (national level) - Hackathon 2.0 Nominated for Undergraduate Research in Abu Dhabi and Dubai competition. Nominated for IEEE Student Research Competition. Nominated for HTC poster competition. Shortlisted in IEEEMadC Competition.