Appeal No. 1996-2240 Application No. 08/105,482 establishing a prima facie case of obviousness in the first instance has not been met. Again, in setting forth the rejection in the first instance, the examiner does not address appellant’s claim limitation respecting a frequency distribution data base. Nor are we persuaded by the examiner’s treatment of this issue in responding to appellant’s arguments. It is apparent from the specification that “conventional mathematical models” are not interchangeable in the claimed method. Nevertheless, the examiner has not explained why frequency distribution probability analysis would have been selected over other models, such as linear regression analysis, stepwise regression analysis, or cluster analysis, which cannot successfully distinguish between disease and non-disease populations. See pages 19 through 21 of the Specification. In our judgment, the reason advanced by the examiner for using frequency distribution analysis in the claimed screening method (“. . . since the frequency distribution shows distinct classifiable differences . . . “) stems from appellant’s description in the specification, and not from the prior art. Accordingly, the rejection of claims 1 through 32 under 35 U.S.C. § 103 as unpatentable over Miyagi, Long and the admitted state of the prior art is reversed. 11Page: Previous 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 NextLast modified: November 3, 2007