-
Example of data sheet being scraped
-
Logo!
Inspiration
Medical professionals are constantly frustrated at the limitless number of medical documents that they have to process and read. One of these documents is the paper medical records which detail outcomes and quality of patient care. Because of the massive amount of non-uniform documents in addition miscategorization of data from current pen to pixel solutions, a need arose to address this issue.
What it does
Our solution, DocToDoc, is a cloud database designed to assist medical professionals in their workflow and modernize data processing in the health industry. DocToDoc starts as an application that captures existing paper medical records and stores it in a cloud database. Through cloud computing, the image is converted to data and data categories are formed into sets. From there, medical facilities, with permission to access the cloud database, can use the data.
Business Model
Although our target and user of our product will be medical professionals, medical professionals nor facilities will be footing the bill for DocToDoc. Because DocToDoc is a database designed to be implemented in (1) medical facilities without proper or effective documentation or (2) medical facilities with existing medical documentation systems, looking for a more efficient solution. For 'low risk' implementation of our system into the health industry, we plan to market our product as a purchasable to health organizations as well as governments that are concerned about the health of their general population. Our solution requires financial support for a cloud database that runs reliably every for every millisecond of life. We estimate the costs to run initially for as low as $3,000 for a country like the Democratic Republic of Congo. We envision the DocToDoc system to be scalable to a larger degree. Our major achievement is a prototype of our own app, capable of connecting to our own optical character recognition engine (OCR), hosted on the cloud.
How we built it
The DocToDoc app is built with Firebase on Android Studio. The computer vision engine of DocToDoc utilizes Google Cloud's Vision API to run our custom-built OCR application that can be adapted to many types of medical record formats.
Challenges we ran into
Our team ran into several challenges throughout the prototyping stage. Our first challenge was manipulating the image to be easily processed by our OCR engine. Another challenge was trying to extract meaningful data from the medical reports and categorizing them. Our last challenge was to ensure a sharp image capture of the document, ensuring maximum quality and readability by our program.
Accomplishments that we're proud of
We are proud that we wrote a unique OCR engine that is designed to extract data from medical records with high precision and designed the system to be easy to use and accessible universally with minimum wifi connection.
What we learned
We learned how to use Google based web tools, i.e Google Cloud Platform Products and Firebase, to build real solutions to address problems in the health field. We also learned the differences used to fund a model using web-based services as compared to the original paper document model.
What's next for DocToDoc
The next step for our prototype is to add Google Cloud machine learning APIs' to recognize categories in different forms and different languages.
Log in or sign up for Devpost to join the conversation.