BMC Medical Informatics and Decision Making

Table 4 Model portability test

From: Medical subdomain classification of clinical notes using a machine learning-based natural language processing approach

From iDASH to MGH					From MGH to iDASH
Subdomain	AUC	Precision	Recall	F1	Subdomain	AUC	Precision	Recall	F1
Cardiology	0.828	0.923	0.715	0.806	Cardiology	0.731	0.829	0.500	0.624
Gastroenterology	0.802	0.396	0.691	0.503	Gastroenterology	0.832	1.000	0.664	0.798
Neurology	0.877	0.745	0.859	0.798	Neurology	0.775	0.902	0.567	0.696
Psychiatry	0.803	0.907	0.613	0.732	Psychiatry	0.941	0.794	0.900	0.844
Pulmonary	0.820	0.197	0.692	0.307	Pulmonary	0.545	1.000	0.089	0.164
Nephrology	0.770	0.573	0.561	0.567	Nephrology	0.634	0.750	0.273	0.400

The performance of using the best interpretable iDASH classifier to classify the medical subdomain of MGH clinical notes, and using the best interpretable MGH model to classify the medical subdomain of iDASH documents

Back to article page

ISSN: 1472-6947

Contact us

General enquiries: journalsubmissions@springernature.com