id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-219880-3wob432t	Ma, Liantao	CovidCare: Transferring Knowledge from Existing EMR to Emerging Epidemic for Interpretable Prognosis	2020-07-17		.txt	text/plain	5686	351	48	Recently, many deep learning-based models have been developed to enable intelligent prognosis by analyzing electronic medical records (EMRs), including mortality prediction [24, 25] , disease diagnosis prediction [22] , and patient phenotype identification [1] . Therefore, for the prognosis of EIDs with limited data, such a research challenge remains: How to make full use of the existing EMR data to learn the robust health status representation, when tackling tasks with different clinical feature sets? In this paper, we propose a novel healthcare predictive approach, CovidCare, based on transfer learning from existing EMR data (i.e., source dataset) to the new dataset (i.e., target dataset) with knowledge distillation. • We propose a transfer-learning-based medical feature embedding approach, CovidCare, to perform clinical prediction for EIDs with limited data .Multi-channel architecture is developed to improve the compatibility across source and target datasets with different feature sets.	./cache/cord-219880-3wob432t.txt	./txt/cord-219880-3wob432t.txt