Neuro-transfer　function　(neuro-TF)　approaches　are　diffusely　applied　in　the　field　of　parametric　modeling　technology,　the　coefficient　of　the　rational　transfer　function　become　more　responsive　to　dealing　with　higher　order　problems;　The　sensitivity　of　the　pole-residue　transfer　function　keeps　at　a　low　level,　but　the　problems　of　noncontinuity　and　non-smooth　of　the　poles/residues　will　appear　when　the　geometric　dimension　changes　greatly.　This　paper　applies　a　hybrid　neuro-TF　to　parameterized　model　of　a　fifth-order　waveguide　filter　to　address　the　above　problem.　From　the　results,　this　method　of　hybrid　neuro-TF　can　obtain　better　improvement　in　accuracy　than　other　methods.　©　2023　IEEE.