id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-027228-s32v6bmd	Subramanian, Vigneshwar	Editorial: Why is modeling COVID-19 so difficult?	2020-06-19		.txt	text/plain	1117	62	51	Disease spread depends heavily on the prevalence of COVID-19, which is not precisely known, and on policy interventions such as social distancing, which are a moving target and not intrinsically measurable. For example, the University of Texas model uses phone geolocation data as a proxy for social distancing and assumes the intervention remains constant across the forecasted time period 5 . Assumptions may also change over time as information emerges and their performance is reassessed; for example, the Columbia model updated contact tracing assumptions to the current parameters to model loosening social distancing restrictions as states reopen 6 . The general workflow involved in developing such a model is as follows: first, the outcome of interest is defined; second, relevant predictors or risk factors are identified; third, the effects of each predictor variable are estimated, for example in a regression analysis; and finally, the model is validated 7 .	./cache/cord-027228-s32v6bmd.txt	./txt/cord-027228-s32v6bmd.txt