In　recent　years,　diagnostic　and　prognostic　methods　of　cancer　have　become　one　of　the　research　focuses　in　clinical　medicine　with　the　increased　risk　of　cancer.　However,　how　to　excavate　the　more　information　from　the　data,　how　to　establish　the　different　math　model,　and　how　to　verify　the　diagnosis　further　are　of　great　significance　for　doctor　to　improve　the　diagnostic　accuracy　and　reduce　the　diagnostic　complexity.　This　paper　collects　and　teases　the　gallbladder　cancer　data　from　the　first　affiliated　hospital　of　medical　college　of　Xi＇an　Jiaotong　University,　China,　which　is　used　to　be　excavated　except　the　known　tumor　markers.　Then,　the　optimal　data　model　is　set　up　using　some　typical　Bayesian　networks.　Finally,　we　improve　the　warning　factors　of　tumor　diagnosis　from　the　known　information　by　putting　importance　calculation　method　into　the　analysis　of　gallbladder　diagnosis.