Low cost Covid-19 preliminary diagnosis utilizing cough samples and keenly intellective deep learning approaches

Abstract

Covid19 an ecumenical pandemic perpetuates to take lakhs of lives and consistently taking its shape as major threat. Skeptically and turmoil in divergent perspectives perpetuate to grow. The most prominent contributing factor to all this is the lack of methodologies to test Covid samples at a more immensely colossal scale. Highly scalable, cost efficacious and flexible diagnosis methodology can contribute greatly towards handling this arduous situation in a more controlled manner. Working towards this the major symptom found among the covid patients is cough. With the avail of Deep learning approaches, this cough is processed to understand the distinctions between the conventional and covid cough. One of the major arduousness to address this quandary is the right amplitude of data to build a deep learning model that can authentically take decisions about the cough recordings. We have extracted some of the recordings from the public platforms and performed deep learning predicated analysis. This gave us the prognostication precision of 94% thus authoritatively mandating a better cough dataset to further carry out the research at a more immensely colossal scale. This paper accommodates as a baseline to cerebrate beyond the customary clinical diagnosis and identify the disease at least in the preliminary in fraction of seconds thus requiring the buildup of covid cough data.