Authors

Inspiration

“Prevention is better than cure” for any virus to progress from endemic to epidemic to pandemic lifecycle stages like attachment, penetration, uncoating, biosynthesis, maturation, Time is of key essence. Whenever a virus enters host body, immune system gets activated and antibodies starts fighting against the virus which ultimately led to change in body vitals like Temperature, cough, shivering etc. These change with respect to each day or time in body vitals along with other health indicators like blood pressure, heart disease, lung disease, blood tests and diabetes becomes a vital clue to detect infection .

Our solution is to perform  online screening for every possible individual using the chatbot on his/her system providing the details on frequent interval ( could be : every alternate day). If symptoms for the effected Persons are getting reflected, our model will find the most probable person or area to be tested.

What it does

Takes all body vital parameters including sore, age, gender, wet cough, Tightness in your chest, Loss of taste and smell, Dizziness, confusion or vertigo, Headache, Muscle aches, Sore throat, runny or blocked nose, cold , low grade fever, temperature  at defined interval of a day / time

  • Cough can be recorded via browser based microphone
  • Record breathe in and out from mouth as deeply as possible five times
  • Record  read aloud any text “ with high pitch”
  • GPS location of the individual
  • Take rapid antibody test results data to check infection
  • Rule based Self-assessment to detect severity of infection
  • In case of individual not able to detect or unsure video call option with doctor to further analyse
  • To detect normal cough and Covid19 or similar other cough an speech model would be built
  • Data collected from GPS location would be used to detect hotspots of specific places to be quarantined
  • Since time is collected places nearby can also mapped
  • If the propagation of virus is human to human, airborne the same would trigger to nearby places that would trigger an alarm to authorities to seal off the complete area it can be plotted over google maps to check progression
Basis the factor provided by individual and time, Markov chain can describe the progression of virus  lifecycle based on chronic and non-chronic diseased individuals 

How I built it

Technologies and Frameworks we have used are Python, R, Angular, REST API, Scikit-Learn, MongoDB, Google Maps, NGINX  Webserver, Tensorflow Using the vital parameters, cough and GPs location, We intent to build an ensemble model for predicting the possibility of individual being infected with the virus.  With GPS location, identifying virus infected hotspots. If a location is not a hotspot, then based on increase in probable cases, we intend to predict the probability and time in which an area can become a hotspot using a markov model.

Challenges I ran into

Limited Cough data availability. Lack of disciplined approach for providing input data to screening engine. Unable to track Protein structure of Rapid Self Mutating Virus.

Accomplishments that I'm proud of

Accuracy of Model

What I learned

Limited Cough data availability. Lack of disciplined approach for providing input data to screening engine. Unable to track Protein structure of Rapid Self Mutating Virus.

What's next for Covid Risk Screening & Tracing (CoRST)

1)Based on the infected area, we can have multiple actions around  food distribution and needy trackers in case of locked down situation. 2) Build the Analytics report based on the data collected

Hackathons

Technologies

angular.js, ec2-servers-on-cloud, google-maps, mongodb, nginx, python, r, rest-api, scikit-learn, tensorflow

Devpost Software Identifier

260380