FinDr

This past Thursday, Jago and his girlfriend were in the hospital due to his girlfriend having blood clots. Blood clots can often lead to serious emergencies, and when asked about this, the nurse simply stated that the assigned doctor would be paged or would be called over the intercom.

However, this system is INCREDIBLY inefficient. Oftentimes, emergencies are a matter of seconds, and waiting for a specific doctor or manually calling them is not an effective solution.

Over 6,000 hospitals still use a system like this. This represents a huge market opportunity. Although microtasks in healthcare have been solved using AI/ML, this is a major operational problem that often gets overlooked because AI/ML focuses on specific diseases.

FinDr can be split into two parts: identifying the problem or condition that a patient may face, and finding a relevant doctor to solve the emergency as effectively as possible.

FinDr utilizes various computer vision-powered methods of tracking patients. We utilized Overshoot for tracking patients using livestreams, and implemented a fine-tuned Sam3 model to analyze the positions of doctors through a network of security cameras (which is low overhead since almost all hospitals have security cameras).

FinDr uses various pathing algorithms to optimize doctor positions and ensure optimal distribution. Doctors can be called based on the relevant condition or personnel, as FinDr also tracks their specialties and the name of the doctor as well.

We believe a system like FinDr is the necessary tool to optimize hospital operations and save hundreds of thousands of lives worldwide.

Built With

• javascript
• overshoot
• python
• sam3
• typescript