Table 9.

Event and attribute classification performance with predicted medication mentions on medications that co-occur with different event and/or attribute labels. 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	0.631	0.679	0.612	0.661	0.814	0.836	0.806	0.843	0.771	0.800
	R	0.612	0.679	0.607	0.661	0.794	0.836	0.804	0.843	0.785	0.800
	F1	0.613	0.679	0.608	0.661	0.802	0.836	0.803	0.843	0.770	0.800
action	P	0.510	0.607	0.432	0.547	0.625	0.779	0.770	0.760	0.903	0.823
	R	0.255	0.306	0.317	0.369	0.548	0.667	0.540	0.685	0.594	0.712
	F1	0.329	0.407	0.361	0.441	0.578	0.718	0.607	0.720	0.674	0.763
actor	P	0.919	0.840	0.496	0.756	0.603	0.857	0.620	0.890	0.462	0.905
	R	0.370	0.568	0.365	0.559	0.499	0.811	0.495	0.802	0.413	0.775
	F1	0.473	0.677	0.420	0.642	0.532	0.833	0.538	0.844	0.437	0.835
temporality	P	0.743	0.696	0.793	0.770	0.801	0.813	0.789	0.840	0.854	0.853
	R	0.483	0.495	0.496	0.514	0.652	0.703	0.693	0.757	0.660	0.730
	F1	0.581	0.579	0.606	0.616	0.715	0.754	0.737	0.796	0.732	0.786
certainty	P	0.607	0.667	0.661	0.732	0.772	0.844	0.806	0.869	0.812	0.884
	R	0.548	0.595	0.479	0.468	0.694	0.730	0.801	0.775	0.764	0.757
	F1	0.569	0.629	0.543	0.571	0.695	0.783	0.795	0.819	0.765	0.816
Overall	P	0.682	0.698	0.599	0.693	0.723	0.826	0.758	0.840	0.761	0.853
	R	0.453	0.529	0.453	0.514	0.637	0.749	0.666	0.772	0.643	0.755
	F1	0.513	0.594	0.508	0.586	0.664	0.785	0.696	0.804	0.675	0.800