The　studies　of　biomedicine　show　that　the　expression　level　of　p53　(tumor　suppressor　gene)　is　related　with　the　tumor　forming.　Researches　of　the　p53　related　signal　transduction　networks　would　provide　new　ideas　for　revealing　the　pathogenesis　of　cancer　or　tumor　and　looking　for　the　treatment　method.　As　the　p53-Mdm2　negative　feedback　network　plays　core　roles,　the　study　of　the　p53-Mdm2　regulation　net　work　is　of　great　significance.　In　this　paper,　based　on　the　gene　expression　time　series　data　of　human　leukemia　cells　after　ionizing　radiation,　the　nonlinear　dynamic　continuous　random　biological　networks　with　time　delay　were　established,　then　continuous-discrete　extended　Kaiman　filter　(EKF)　algorithm　was　used　to　simulate　the　dynamic　regulation　of　p53　and　Mdm2.　Meanwhile,　the　accuracy　of　the　model　established　in　this　paper　was　validated　and　we　found　that　the　error　rate　of　the　model　was　only　0.　85%.　Results　showed　that　the　algorithm　thatis　convergent　could　predict　gene　expression　level　at　any　time　and　simulate　accurately　the　damped　oscillation　process　of　p53-Mdm2　regulatory　networks　after　ionizing　radiation.　The　response　process　was　consistent　with　biological　experiment　results.　The　algorithm　proposed　in　this　paper　provided　an　effective　method　for　biological　experiment　modeling,　and　supplied　the　foundation　for　the　research　of　system　dynamics.