A Diverse Stochastic Search Algorithm for Combination Therapeutics

Table 1

Summary of the results for seven synthetic and two experiment based examples. The table illustrates that the cost for our proposed algorithm is significantly lower than the ARU algorithm [15] which has been shown to be efficient than other existing algorithms. Furthermore, the success rate for our algorithm is also better than the ARU algorithm.

	Number of points with ≥0.95×Maxefficacy	Search grid size	ARU [15] cost	ARU [15] Success Rate	DSS cost	DSS STD	DSS worst case	DSS Success Rate	Initial LHC points
Synthetic examples
2 DeJong	2	21^2	46.2	99%	16.0	7.99	48	100%	5
3a	4	11^3	74	100%	24.7	11.73	72	100%	10
3b	1	11^3	79.4	100%	52.9	32.20	149	100%	10
4a	1	11^4	136.8	100%	65.3	14.11	106	100%	40
4a	1	11^4	136.8	100%	50.7	21.81	159	100%	10
4b	12	11^4	91.6	100%	52.7	8.80	85	100%	40
4b	12	11^4	91.6	100%	28.3	9.17	57	100%	10
5a	4	11^5	80.6	100%	79.3	23.25	157	100%	40
5a	4	11^5	80.6	100%	61.8	27.58	176	100%	10
5b	8	11^5	216.8	100%	159.5	90.51	402	100%	40
5b	8	11^5	216.8	100%	194.2	150.15	647	100%	10
Experiment based examples
Bacterial inhibition [10]	34	9^2	4.8	100%	1.85	0.78	3	100%	3
Lung cancer [16]	7	10^2	12.4	98%	5.97	4.74	23	100%	3