Machine Learning and Data Mining Methods in Early Detection of Stomach Cancer Risk

• Published in: 淡江理工學刊 Vol. 24; no. 1; pp. 1 - 8
• Main Authors: Rejaul Islam Royel, Ajmanur Jaman, Fuyad Al Masud, Arzo Ahmed, Abdul Muyeed, Kawsar Ahmed
• Format: Journal Article
• Language: English
• Published: 淡江大學 01.01.2021
• Subjects: Correlation; Stomach Cancer in Bangladesh; Feature Selection Method; Risk algorithm; Preoperative Risk Factors
• ISSN: 2708-9967
• DOI: 10.6180/jase.202102_24(1).0001

Background: In point of mortality rate, stomach cancer is the fifth leading cancer. There are some risk factors of stomach cancer those are varied with country to country and associated with urbanization and economic development. Diagnosis of stomach cancer is a difficult task, only about 10% of people are diagnosed while it's still in the initial stage. The main objective of this research is to design a tool for early detection of stomach cancer risk level. Methodology: Firstly, feature selection techniques are applied to filter the collected data. After that, the best rules technique is used to check the correlations of risk factors with stomach cancer. Besides, the visual relationship among factors and selected cancer are also exhibited. Then, the score is assigned for each factor according to the impact of risk on stomach cancer patients. Finally, the stomach cancer risk level prediction tool is designed. Results: After the experiment of 300 subjects' records (150 are affected and 150 are non-affected) with 32 risk factors, we have received 18 significant-top risk factors of stomach cancer. Abdominal Pain, Nausea, Skin Color Turn into pale are respectively found top risk factors of stomach cancer. Furthermore, some other factors related to socio-economic conditions are also indicated to have stomach cancer. Conclusion: In conclusion, this study will be helpful to early detection of stomach cancer risk level and to increase the awareness among the people of Bangladesh as well as the rest of the world.