Decision tree-based method for integrating gene expression, demographic, and clinical data to determine disease endotypes

Table 7 Accuracy per Method/Weighting scheme

Method	Classification %	Groups Asthmatic/Non-Asthmatic
33/33/33	65	4/5
20/40/40	66	3/4
40/20/40	67	6/6
50/50/0	56	2/4
30/60/10	61	3/3
60/30/10	63	4/8
40/40/20	66	4/8
Adaptive	60	1/3
Decision	78	4/4

The weighting schemes are shown as Gene Expression Domain Weighting/Clinical Covariate Domain Weighting/Indicators of Disease Status Domain Weighting. Column 2 shows the percentage of subjects correctly classified by asthma status after assigning asthma status based on the majority of subjects in the group. Column 3 shows the number of asthmatic and non-asthmatics groups respectively using this definition. Entries in bold represent those methods showing at least 65% accuracy. Information on the classification accuracy of the individual asthmatic clusters for these methods is shown in Table 8.

ISSN: 1752-0509