Optimizing the maximum reported cluster size in the spatial scan statistic for survival data

Lee, Sujee; Moon, Jisu; Jung, Inkyung

doi:10.1186/s12942-021-00286-w

International Journal of Health Geographics

Table 4 Simulation results for cluster model D (one elliptical cluster, 10% of total area) with a mean survival time of 5

From: Optimizing the maximum reported cluster size in the spatial scan statistic for survival data

	% of cens		Maximum reported cluster size (MRCS)																Default Setting
	% of cens		3%	4%	5%	6%	8%	10%	12%	15%	20%	25%	30%	35%	40%	45%	50%	Overall	Default Setting
Circular window	10%	Frequency	14	3	8	196	256	82	20	41	84	57	36	17	20	6	5		845
		Sensitivity	0.017	0.278	0.500	0.669	0.650	0.689	0.700	0.797	0.863	0.974	0.986	0.971	0.992	1.000	1.000	0.730	0.617
		PPV	1.000	0.778	0.800	0.970	0.891	0.697	0.553	0.509	0.461	0.410	0.328	0.270	0.239	0.200	0.172	0.728	0.726
	30%	Frequency	6	3	10	240	365	92	7	35	63	44	11	11	4	5	2		898
		Sensitivity	0.011	0.333	0.617	0.653	0.655	0.714	0.714	0.805	0.833	0.920	0.985	1.000	0.958	0.967	1.000	0.699	0.629
		PPV	0.833	1.000	0.860	0.989	0.910	0.678	0.568	0.501	0.453	0.385	0.327	0.270	0.212	0.192	0.182	0.807	0.802
	50%	Frequency	3	0	8	87	415	49	7	37	90	84	9	16	16	7	19		847
		Sensitivity	0.003	–	0.667	0.634	0.659	0.660	0.619	0.766	0.782	0.865	0.926	0.979	0.990	0.929	0.974	0.716	0.612
		PPV	0.667	–	0.767	0.966	0.867	0.694	0.531	0.503	0.425	0.353	0.296	0.270	0.227	0.196	0.176	0.698	0.691
	70%	Frequency	0	0	2	25	142	9	1	7	25	32	99	32	5	77	40		496
		Sensitivity	-	–	0.500	0.640	0.612	0.648	0.333	0.762	0.807	0.823	0.978	0.995	0.967	0.903	0.883	0.807	0.426
		PPV	-	–	1.000	0.960	0.768	0.707	0.286	0.491	0.444	0.325	0.329	0.265	0.228	0.191	0.171	0.464	0.437
Elliptical window	10%	Frequency	1	5	14	28	67	426	140	101	64	34	15	11	8	4	2		920
		Sensitivity	0.004	0.300	0.488	0.667	0.704	0.974	0.973	0.965	0.961	0.976	0.978	0.985	0.958	1.000	1.000	0.931	0.857
		PPV	1.000	0.800	0.976	0.966	0.945	0.953	0.817	0.657	0.525	0.404	0.327	0.278	0.229	0.205	0.191	0.820	0.820
	30%	Frequency	2	2	9	57	68	460	160	96	52	17	8	3	3	2	2		941
		Sensitivity	0.006	0.333	0.556	0.670	0.696	0.959	0.959	0.953	0.910	0.951	0.958	0.944	0.944	0.917	1.000	0.912	0.858
		PPV	0.500	1.000	0.944	0.994	0.973	0.937	0.819	0.663	0.493	0.405	0.305	0.257	0.226	0.174	0.188	0.847	0.848
	50%	Frequency	0	3	6	3	54	340	248	134	54	37	11	6	12	4	12		924
		Sensitivity	–	0.222	0.472	0.722	0.735	0.960	0.962	0.969	0.880	0.896	0.864	0.972	0.958	0.958	0.972	0.934	0.864
		PPV	–	0.667	0.944	0.875	0.944	0.939	0.825	0.674	0.473	0.369	0.279	0.265	0.230	0.187	0.182	0.785	0.783
	70%	Frequency	0	1	12	13	105	301	18	38	21	41	39	30	60	20	34		733
		Sensitivity	–	0.333	0.472	0.500	0.730	0.976	0.759	0.886	0.746	0.850	0.897	0.956	0.986	0.942	0.966	0.893	0.657
		PPV	–	1.000	0.944	1.000	0.981	0.924	0.667	0.594	0.416	0.349	0.290	0.265	0.241	0.204	0.182	0.692	0.690

% of cens, percentage of censoring; PPV, positive predictive value
Cells most often selected as the optimal MRCS are shown in italics

Back to article page

ISSN: 1476-072X

Contact us

Submission enquiries: journalsubmissions@springernature.com