The COVID-19 pandemic is turning out be one of the worst pandemics of the modern times. COVID-19 will certainly not be the last pandemic the world sees unless proper centralised privacy-protected global disease monitoring systems are employed. A Centralised disease monitoring system if achieved has the potential to not only trap future pandemics to every early stages it can also provide insights into diseases themselves.
Coronavirus outbreak could have been stopped at early stages if there was a system to detect anomalies in the disease patterns.

What it does

The web-app takes the recorded doctor-patient conversation and converts into text. It then extracts the common symptom terms and sends it to the disc for being recorded. The symptoms recorded are grouped into zones and check- marked every 6 hours. It supports a tabular representation of checkpoint wise symptoms. These symptoms can be then used by data-scientists to understand and diagnose abnormal disease patterns.

How I built it

The backend was build using flask- a lightweight micro-framework. The experimental front end was developed using HTML and CSS. This front end can be easily replaced with an Android App easily as the functionalities of the backend are factored into simple APIs.

Challenges I ran into

For the purposes of this web app the world was divided into 720*360 = 259200 zones. Using a database to extract this much information would be a time taking task .So instead of using a database all these zones were serialized into memory using python and accessed form the disc which lead to more efficient data manipulation.

Accomplishments that I'm proud of

The best part of this app is that it does not require any major external work from the doctor. It is not very difficult to scale to even world-wide scenarios as it does not require any reskilling of the medical staff. The making the working as invisible would the biggest accomplishment.

What's next for Real  time centralised web-based disease monitoring system.

This proof-of-concept uses simple regular expressions for symptom extraction. I believe that even simple regular expressions would not work as bad as you would expect as most of the medical conversations are to the point and even simple REs would work well,  however deep learning based NLP algorithms will be able to extract more complex symptoms form the text .

Try It out



flask, python, voice-recognition

Devpost Software Identifier