-
CardioScope logo
Inspiration
Around the world each year, tens of millions of lives are sadly taken away short by heart disease. This global issue is especially prevalent in wealthier nations and areas with higher standards of living, as wealth index and body mass index are positively correlated. Unfortunately, most types of cardiac disease are incurable, and the only solution is to mitigate symptoms through treatment. The most common method of diagnosing heart disease is through doctor visits, where they take echocardiogram readings, blood tests, etc., to find out if the patient is at risk or has heart disease. However, many people delay their doctor visits for whatever reason - financial issues, timing issues, laziness - anything. This is where CardioScope comes in, providing a simple and cost-effective solution that can predict heart disease based on very common and accessible data.
What it does
CardioScope is a program that takes in a wide variety of data in .csv files, which can be taken from your Apple Watch, which tracks your heart rate and electrocardiographical data using the Health and Heart apps. This data can then be exported to your iPhone/iPad/iPod with the simple click of a button already integrated into iOS. This data can then be inputted into our program, which uses our datasets and machine learning models to generate an accurate prediction of whether a person is healthy, or has conditions such as coronary artery disease, aortic disease, or arrhythmia.
How we built it
To build CardioScope's heart disease tracking technology, our team followed a specific set of steps. We gathered data from echocardiogram recordings, heart rate data sent from any model of Apple Watch. Next, we preprocessed and cleaned the data to remove any noise or inconsistencies. After that, we developed machine learning models to make predictions based on the input data, selecting appropriate algorithms, tuning hyperparameters, and testing the models on a validation dataset. We then integrated the models into a web application that could receive input data, process it using the models, and display results to users, while ensuring the application was secure and scalable. Finally, we thoroughly tested and deployed the application.
Challenges we ran into
One specific challenge we encountered while integrating Apple Watch data into CardioScope's web application was related to the different versions of the WatchOS operating system. We found that some users were experiencing issues with data synchronization when they upgraded to the latest version of WatchOS, which could impact the accuracy and reliability of our machine learning models. To solve this problem, our team worked closely with Apple's developer resources to identify and fix the issue. We updated our algorithms and data processing pipeline to handle the changes in data format and synchronization, and we also implemented additional error handling and data validation checks to ensure that the data received from the Apple Watch was accurate and reliable.
Accomplishments that we're proud of
We are proud of the fact that we were able to work and create a program using Apple devices, as well as using ML models, things that some of our team members haven't done before. We are also proud of the fact that we were able to come up with a full-fledged program prototype after we spent a long portion of the day stuck planning what we were going to do.
What we learned
We learned how to integrate data from Apple devices, as our group had only worked in mobile apps using Android Studio and devices before, and using Xcode was a first step for everyone on our team. The project was also a learning experience for some of our teammates, who had only worked with back-end aspects in the past, now learning how to work with front-end elements.
What's next for CardioScope
We look forward to improving our accuracy by adding more data as well as improving our current machine-learning models. We also plan to add an interactive map feature where the user can locate nearby cardiologists and be recommended the best one for them based on factors such as income, location, insurance, etc.
Log in or sign up for Devpost to join the conversation.