The problem MediGuide solves

In rural hospitals, there is acute lack of medical expertise. In contrast, busy urban hospitals face problem of limited medical personnel dealing with multitude of patients. Emergency scenarios can occur in non-peak schedules when medical expertise is not available. This leads to critical care being burdened on the shoulders of very few doctors, residents and interns.
To solve the so forth mentioned problems, we propose MediGuide, a deep learning fueled web-platform to retrieve medical history of the previous cases which are similar to the current case being diagnosed. The proposed framework operates over different medical modalities such as Medical Images and Physiological Measurement. In this work, we have incorporated Brain-Tumor MRI and ECG Arrhythmia retrieval. Nevertheless, the system is generalized for any form of medical signal. Retrieving similar results and corresponding case history can help doctors make medical inference fast and more accurate. We are the first Multi-Modality Medical Retrieval Framework as well first Physiological Signal (ECG, EEG, PPG etc) retrieval framework.

Challenges we ran into

• Development of Accurate Machine Learning Models: We developed doamin-specific architectures to achieve high
  performance.
• Integrating front-end with the back-end
• Deploying models: We faced the biggest challenge in deploying the ML-Model in the backend, to this front, after hours of trial and error we were finally able to get this done.