MedApp is a comprehensive health diary application that allows you to focus on your health by leveraging the latest in Artificial Intelligence to surface relevant insights about your conditions, symptoms, medications, and diet.
We wanted to work on something that challenged us technically, but also something that would have a positive impact on people. With the prevalence of health logging apps, we identified a gap that we could take a stab at.
At the surface, our app resembles the myriad of health logging apps that allows users with chronic conditions to keep track of their medications, symptoms, food intake, etc. However, we take it one step further by identifying latent factors that exacerbate or ameliorate their symptoms. This allows the users of the app to have a higher quality of life and make more informed decisions about their health.
We developed an android application that allowed users to seamlessly enter information about their health which they could access to review their history but also integrate information about insights gleaned from their data. To generate insights, we used anonymized health data aggregated from open source online sources to identify contributions of age, sex, medications, diet, and diseases on commonly reported symptoms in health diary apps. We used a Bayes State Space Time Series model to identify contribution of latent variables to symptom severity. Additionally, we were able to build a classification model that identified class of foods associated with least symptoms given disease and medication.
Lack of sleep being the obvious challenge, we encountered some roadblocks during pre-processing the health data and dealing with incomplete time series, missing data, etc. We also encountered challenges while trying to load the custom machine learning model within the app in order to make predictions.
We are proud of the layout and interface we were able to create for our application. We believe it offers a really intuitive experience for the users not only to record their health status but also glean insights to improve their health. Given the limited time, we are also proud about the analysis pipeline we were able to establish to provide insights to our users.
We were exposed to and got experience building android applications and also in analyzing real world health data along with time series analysis.
Given more time, we would love to fully integrate our models within the app and validate our approach with real world data. It would also be nice to adopt some more complex models/analysis to increase the quantity and quality of insights we can provide.
JAVA/Android Studio, HTML, Python
Google Home Mini
$100 Amazon Gift Cards
Blu R2 Plus Smartphones
Misfit Shine 2