Table 8.

Event and attribute classification performance with predicted medication mentions on medications that co-occur. We used the Clinical BERT model to extract the medications.

Model	Metric	MedSingleTask		MedMultiTask		MedSpan		MedIdentifiers		MedIDTyped
		Macro	Micro	Macro	Micro	Macro	Micro	Macro	Micro	Macro	Micro
event	P	0.729	0.862	0.754	0.876	0.820	0.899	0.836	0.911	0.768	0.886
	R	0.643	0.862	0.704	0.876	0.795	0.899	0.813	0.911	0.810	0.886
	F1	0.679	0.862	0.727	0.876	0.807	0.899	0.824	0.911	0.780	0.886
action	P	0.577	0.676	0.684	0.680	0.830	0.787	0.774	0.780	0.879	0.800
	R	0.326	0.391	0.437	0.531	0.608	0.714	0.590	0.703	0.623	0.729
	F1	0.401	0.495	0.506	0.596	0.667	0.749	0.655	0.740	0.704	0.763
actor	P	0.873	0.747	0.515	0.810	0.587	0.832	0.602	0.861	0.430	0.851
	R	0.371	0.615	0.374	0.620	0.457	0.776	0.457	0.776	0.407	0.771
	F1	0.436	0.674	0.431	0.702	0.490	0.803	0.497	0.816	0.418	0.809
temporality	P	0.748	0.709	0.839	0.824	0.819	0.810	0.794	0.832	0.836	0.822
	R	0.531	0.557	0.562	0.609	0.671	0.708	0.702	0.750	0.673	0.745
	F1	0.617	0.624	0.667	0.701	0.735	0.756	0.745	0.789	0.736	0.781
certainty	P	0.624	0.704	0.704	0.812	0.787	0.841	0.796	0.861	0.764	0.851
	R	0.611	0.656	0.537	0.583	0.701	0.745	0.798	0.776	0.728	0.771
	F1	0.605	0.679	0.597	0.679	0.710	0.790	0.792	0.816	0.737	0.809
Overall Avg	P	0.710	0.739	0.699	0.800	0.769	0.834	0.761	0.849	0.735	0.842
	R	0.497	0.616	0.523	0.644	0.646	0.768	0.672	0.783	0.648	0.780
	F1	0.548	0.667	0.586	0.711	0.682	0.799	0.703	0.815	0.675	0.810