Diagnosis of prostate cancer and predicting the probability of suffering the disease in workers

Abstract

Introduction: Diagnosis of various diseases in medicine is one of the areachr('39')s most widely used data mining in recent years and many researches have been done about it. In this study, the diagnosis of prostate cancer using fuzzy system was assessed. The goal was to diagnose the prostate cancer and to predict the possibility of suffering from the disease. Methods: In the proposed method, at first, based on available dataset, pre-processing and clustering operations were carried out. Then a zero-order Sugeno fuzzy system was designed for prediction. Each cluster, as the first item of a fuzzy rule, was considered and out of a rule, percentage of disease possibility in each cluster was considered. For each new sample, the membership degree to the each cluster was computed and then by combining outputs from each rule, possibility of disease in the sample was predicted. Finally, by having possibility and threshold for possibility, having or not having the disease for desired sample was diagnosed. Results: The results showed that the system has good accuracy in predicting the possibility of disease. Conclusion: The results of this study can be used to predict the risk of prostate cancer in young workers according to different jobs they are employed and the amount of exposure to risk factors in each job. If this possibility is high, they are known as the person at risk and in some cases there may need to change higher jobs.

Keywords

• prediction
• prostate cancer
• fuzzy systems