Extending electronic medical records vector models with knowledge graphs to improve hospitalization prediction

Table 8 PR_avg,RE_avg,F1_avg,AUC_avg and their standard error variations computed between each folds for the different vector sets considered on the balanced dataset DS_B under logisitc regression

Features set	PR_avg	RE_avg	F1_avg	AUC_avg	STD(PR)	STD(RE)	STD(F1)	STD(AUC)
baseline	0.8786	0.8236	0.8490	0.8551	0.0473	0.0484	0.0353	0.0334
+t	0.8819	0.8306	0.8546	0.8600	0.0439	0.0344	0.0283	0.0280
+c₁	0.8798	0.8152	0.8453	0.8523	0.0435	0.0432	0.0323	0.0309
+c₁₋₂	0.8775	0.8278	0.8511	0.8565	0.0442	0.0400	0.0320	0.0311
+c₂	0.8795	0.8250	0.8508	0.8565	0.0442	0.0335	0.0315	0.0309
+d_prevent	0.8756	0.8235	0.8478	0.8537	0.0420	0.0457	0.0322	0.0302
+d_treat	0.8740	0.8251	0.8482	0.8538	0.0403	0.0353	0.0321	0.0309
+d_CI	0.8721	0.8236	0.8462	0.8517	0.0506	0.0449	0.0375	0.0366
+wa	0.8816	0.8306	0.8545	0.8600	0.0451	0.0443	0.0348	0.0334
+wi	0.8766	0.8264	0.8498	0.8559	0.0377	0.0506	0.0348	0.0321
+wm	0.8730	0.8221	0.8458	0.8516	0.0436	0.0430	0.0320	0.0311
+s	0.8766	0.8235	0.8484	0.8544	0.0442	0.0457	0.0357	0.0337
+s∗	0.8799	0.8264	0.8502	0.8572	0.0446	0.0446	0.0341	0.0325
+s∗T	0.8755	0.8025	0.8375	0.8466	0.0256	0.0634	0.0405	0.0329
+s∗∩	0.8800	0.8094	0.8420	0.8507	0.0269	0.0597	0.0368	0.0322
+s∗∪	0.8734	0.8177	0.8433	0.8508	0.0282	0.0633	0.0399	0.0337
+s∗m	0.8929	0.8376	0.8639	0.8642	0.0259	0.0398	0.0280	0.0258
+sm	0.9001	0.8404	0.8686	0.8744	0.0267	0.0431	0.0287	0.0261
+sm∩	0.8966	0.8389	0.8660	0.8717	0.0349	0.0427	0.0296	0.0277
+sm∪	0.9008	0.8445	0.8712	0.8765	0.0283	0.0378	0.0257	0.0240

ISSN: 2041-1480