id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-309096-vwbpkpxd	Magdon-Ismail, Malik	Machine Learning the Phenomenology of COVID-19 From Early Infection Dynamics	2020-03-20		.txt	text/plain	4881	412	67	We present a robust data-driven machine learning analysis of the COVID-19 pandemic from its early infection dynamics, specifically infection counts over time. We also follow a data-driven machine learning approach to understand early dynamics of COVID-19 on the first 54 days of US confirmed infection reports (downloadable from the European Center For Disease Control). β, asymptomatic infectious force governing exponential spread γ, virulence, the fraction of mild cases that become serious later k, lag time for mild infection to become serious (an incubation time) M 0 , Unconfirmed mild asymptomatic infections at time 0 Figure 1 are the model predictions (blue envelope) and the red circles are the observed infection counts. Our results demonstrate the effectiveness of simple robust models for predicting pandemic dynamics from early data. From this solution as a starting point, we can further optimize γ, β using a gradient descent which minimizes an error-measure that captures how well the parameters β, γ, k, M 0 reproduce the observed dynamics in Figure 2 , see for example Abu-Mostafa et al.	./cache/cord-309096-vwbpkpxd.txt	./txt/cord-309096-vwbpkpxd.txt