An end-to-end hybrid algorithm for automated medication discrepancy detection

Table 1 Descriptive statistics of the annotated clinical notes

Statistics	Overview notes			Discharge summaries			Overall
Encounters	975			168			975
Patients	271			112			271
Notes	4025			300			4325
Category	Discr	Match	All	Discr	Match	All	Discr	Match	All
Medication name	1116	3873	4989	24	2935	2959	1140	6808	7968
Amount	169	1270	1439	47	1937	1984	216	3207	3423
Dosage	106	857	963	26	711	737	132	1568	1700
Duration	4	16	20	0	70	70	4	86	90
Form	36	1246	1282	6	2434	2440	42	3680	3722
Frequency	410	2742	3152	38	2635	2673	448	5377	5825
Route	48	443	491	42	2836	2878	90	3279	3369
Strength	142	1281	1423	4	2092	2096	146	3373	3519
Overall	2031	11728	13759	187	15650	15837	2218	27378	29596

“Discr” column shows the number of discrepant entities and “Match” the number of matched entities.

ISSN: 1472-6947