Table 5.

Empirical probabilities of possessing at least latent features $k_1$ and $k_2$ given that $k_1$ is active.

$k_1$	$k_2$
	1 (%)	2 (%)	3 (%)	4 (%)	5 (%)	6 (%)	7 (%)	8 (%)	9 (%)	10 (%)	11 (%)	12 (%)	13 (%)	14 (%)
1	100.00	2.30	1.89	1.87	1.65	1.53	1.34	1.29	1.24	1.13	1.12	1.11	1.07	1.01
2	100.00	100.00	3.55	4.64	2.88	4.17	2.33	4.72	2.83	1.37	1.94	1.71	5.02	3.53
3	100.00	4.31	100.00	3.44	2.08	2.32	2.08	2.20	1.72	1.51	1.81	1.75	2.56	1.27
4	100.00	5.71	3.48	100.00	1.56	2.96	2.53	4.31	2.93	1.83	2.38	3.08	2.81	2.14
5	100.00	4.00	2.38	1.76	100.00	2.10	2.07	2.34	2.79	1.79	1.86	2.48	1.69	1.83
6	100.00	6.28	2.88	3.63	2.28	100.00	2.36	3.74	2.43	1.57	4.75	2.24	2.69	2.88
7	100.00	4.01	2.95	3.54	2.56	2.69	100.00	1.84	2.01	1.79	2.05	1.96	1.75	1.20
8	100.00	8.39	3.22	6.23	3.00	4.42	1.90	100.00	3.22	2.47	2.65	2.52	2.61	2.08
9	100.00	5.25	2.63	4.42	3.73	3.00	2.17	3.36	100.00	3.27	2.30	3.27	2.21	1.75
10	100.00	2.78	2.53	3.03	2.63	2.12	2.12	2.83	3.59	100.00	2.27	2.53	1.87	1.92
11	100.00	3.97	3.05	3.97	2.75	6.46	2.44	3.05	2.54	2.29	100.00	2.95	2.70	2.34
12	100.00	3.56	2.99	5.21	3.71	3.09	2.37	2.94	3.66	2.58	2.99	100.00	2.27	2.53
13	100.00	10.80	4.54	4.92	2.62	3.85	2.19	3.15	2.57	1.98	2.83	2.35	100.00	1.92
14	100.00	8.00	2.37	3.94	2.99	4.34	1.58	2.65	2.14	2.14	2.59	2.76	2.03	100.00

To compute the empirical probabilities, we looked at the number of patients that had latent feature $k_1$ active and counted how many of them also have latent feature $k_2$ active.