Table 5.

Summary of the results of absolute (A) models

Missing baseline data	Model name	Training data (n)	CV/test	r	r 2	P value	MAE
None	AG	13 936	CV	0.75	0.56	0.0001	0.65
			test	0.74	0.55	0.0001	0.69
Genotype	ANG	56 717	CV	0.70	0.49	0.0001	0.71
			test	0.68	0.46	0.0001	0.73
Time on therapy	AG(−ToT)	19 080	CV	0.74	0.55	0.0001	0.65
			test	0.72	0.52	0.0001	0.70
Genotype/time on therapy	ANG(−ToT)	50 270	CV	0.69	0.48	0.0001	0.72
			test	0.67	0.45	0.0001	0.74
Genotype/CD4 count	ANG(−CD4)	62 940	CV	0.70	0.49	0.0001	0.71
			test	0.69	0.48	0.0001	0.75
Genotype/time on therapy/CD4 count	ANG(−ToT −CD4)	56 717	CV	0.68	0.46	0.0001	0.74
			test	0.66	0.44	0.0001	0.75

CV, cross-validation; test, independent testing; MAE, mean absolute error - the difference between predicted and actual change in plasma viral load (log₁₀ copies HIV RNA/mL).