Assessing misdiagnosis of relapse in patients with gastric cancer in Iran cancer institute based on a hidden Markov multi-state model.

摘要: Background: Accurate assessment of disease progression requires proper understanding natural process which is often hidden and unobservable. For this purpose, status should be clearly detected. But in most diseases it not possible to detect such status. This study, therefore, aims present a model both investigates the unobservable considers error probability diagnosis states. Materials Methods: Data from 330 patients with gastric cancer undergoing surgery at Iran Cancer Institute 1995 1999 were analyzed. Moreover, estimate assess effect demographic, diagnostic clinical factors as well medical post-surgical variables on transition rates misdiagnosis relapse, Markov multi-state was employed. Results: Classification errors alive state without relapse (e 21 ) 12 0.22 (95% CI: 0.04-0.63) 0.02 0.00-0.09), respectively. Only age number renewed treatments affected relapse. In addition, patient distant metastasis among affecting occurrence (state1"state2) while type extent had significant death hazard (state2"state3)and (state2"state3). Conclusions: A provides possibility estimating classification between different states disease. based model, can identified researchers helped mechanisms error.