A Robust Statistical CNN CTC Based AI Model for Tracking and Monitoring Covid-19

EasyChair Preprint 3683

Abstract

The world today battles the new coronavirus (COVID-19) spread that can lead to a major endemic or pandemic outbreak with devastating outcomes which has overturned all our usual calculus seemingly. Due to the unavailability of a vaccine and the high degree of contagiousness, social distancing is perhaps the only way left to battle the virus. The tracking system is still a challenge for monitoring the COVID-19 affected person and the community spread. This leads to an epidemic situation, where the number of secondary infections keep going higher and higher which intern leads to the difficulty of tracking and treating each infected person. An AI based mathematical modelling is proposed for tracking and monitoring the pandemic outbreak. A GPS tracking system, geo location using Fit bit and wifi tracking modules are developed to locate the secondary infections and the carriers. Calculation of real time co-ordinates and the conditional probabilities with respect to time is done followed by the classification based on risk using CNN CTC (Connectionist Temporal Classification) model, a type of neural network which is used to train the recurrent process. Monitoring the affected person using sensors like thermal sensors and a tracking chip inbuilt in a wristband for real time updates. A comparative study is done on various Classification models. Therefore, the system provides a robust model with 96.5% accuracy to combat this pandemic situation.

Keyphrases: CNN, COVID-19, Connectionist Temporal Classification (CTC), Fit bit, GPS, Thermal Sensors

@booklet{EasyChair:3683,
  author    = {Dr.Ballika J Chelliah and Arunkumar Sadayapillai and Prabavathi Raman and Aarthi Babu},
  title     = {A Robust Statistical CNN CTC Based AI Model for Tracking and Monitoring Covid-19},
  howpublished = {EasyChair Preprint 3683},
  year      = {EasyChair, 2020}}