CorCheck

Problem: When an ambulance races through the city, carrying a patient with potential cardiac issues, the big question arises – should this patient be rushed to the general hospital or directly to a specialized cardiology clinic? Making this decision swiftly and accurately is vital, yet incredibly challenging.

Solution: We introduce CorSort, an innovative neural network designed to analyze EKG data in real-time, straight from the ambulance. Our technology swiftly categorizes patients into three critical pathways: direct to a cardiology clinic for specialized care, to a general hospital for broader medical attention, or in certain cases, recommending care at home.

Impact: The right decision not only saves precious time but also ensures that patients receive the most appropriate care without overwhelming hospital resources. With CorSort, we aim to enhance the efficiency of emergency medical services, reduce the strain on healthcare systems, and most importantly, save lives.

Feasibility & financials: CorSort's value lies in reducing costs from wrong ambulance routing—like unnecessary treatments and transfers—by ensuring correct patient placement. This efficiency lowers healthcare expenses and optimizes resource use. If savings from right decisions outweigh the costs of errors, CorSort proves financially feasible, offering significant ROI and enhancing emergency care.

What is new about your solution?: Our solution introduces a groundbreaking neural network model, uniquely crafted by integrating three advanced approaches: Triplet Loss, Autoencoders, and diverse yet complementary techniques. This fusion not only enhances the accuracy of our model but also ensures robust and efficient patient categorization based on EKG data.

What you have built at the hackathon - text explanation + code (e.g. GitHub link): At the hackathon, we developed an innovative neural network model, CorSort, aimed at analyzing EKG data from ambulances to determine the most appropriate care pathway for patients. CorSort uniquely blends Triplet Loss functions, Autoencoders, and a variety of advanced neural network techniques to categorize patients into 'Stay Home', 'Normal Hospital', or 'Cardiology Clinic' https://github.com/CaptainTrojan/hackhealth2023

What you had before the hackathon, please mention open source as well: https://github.com/CaptainTrojan/electrocardioguard

What comes next and what you wish to achieve: Next, we aim to deploy CorSort for real-world testing, focusing on maximizing its accuracy. Our primary goal is aiding patients by providing rapid, accurate EKG analysis for better emergency response. Concurrently, we plan to publish a scientific paper detailing CorSort's innovation and impact, contributing to medical AI research and practice.

Video: https://www.youtube.com/embed/TiFiLMHiq6s