Infection status outcome, machine learning method and virus type interact to affect the optimised prediction of hepatitis virus immunoassay results from routine pathology laboratory assays in unbalanced data

Table 4 Analysis of variance of mean accuracy rates for a four-factor experiment

Source	SS	df	MS	F	p
Method	28.015	2	14.008	0.488	0.620
Pre-processing	0.967	3	0.322	0.011	0.998
Virus	44.815	1	44.815	1.560	0.224
Outcome	927.169	1	927.169	32.279	0.000 (*)
Method.Outcome	2909.082	2	1454.541	50.640	0.000 (*)
Method.Pre-processing	0.863	6	0.144	0.005	1.000
Method.Virus	42.649	2	21.324	0.742	0.487
Pre-processing.Outcome	8.436	3	2.812	0.098	0.960
Virus.Outcome	922.604	1	922.604	32.120	0.000 (*)
Pre-processing.Virus	0.301	2	0.100	0.003	1.000

The experiment examines interactions affecting the prediction of HBSA and HepC immunoassay outcome.
(*) = Significant at 0.001 level.
Method = basic single, basic multiple, majority multiple or clear negative.
Pre-processing = none, log, scale, or scale-log.
Virus = Hepatitis B or Hepatitis C.
Outcome = positive or negative.
Method.Outcome = the interaction between method and outcome; other interactions between pairs of variables to be interpreted similarly.

ISSN: 1471-2105