Predictive Diagnosis of Cancer using Machine Learning

Abstract

In Today’s world, Technology has revolutionized almost every aspect. However,Despite the advancements various diseases are still in the rise. Among them , Cancer stands as one of the major cause of death and accounts for about 9.6 million deaths worldwide . However, current evidence suggests that by the introduction of a proactive system which incorporates the avoidance and modification of key risk factors and by introducing a mechanism which enables detecting and predicting cancer at the earliest stage accurately would pull down the cancer death percentage by 30-50%. Technology-enabled smart healthcare is no longer a flight of fancy. However, The required facility for diagnosing cancer accurately and at the earliest stage using the results of biopsy are not available to  all general hospitals. Identifying and diagnosing cancer at the earliest stage is crucial as the possibility of cancer spreading increases. Therefore, A  computerized system which identifies cancer at the earliest stage with minimal time with greatest accuracy and which reduces cancer recurrence and mortality has to be developed. This paper concentrates and summarises the different machine learning algorithms which may be implied in cancer diagnosis to improve the accuracy of the diagnosis and identification.

Objective: Though numerous data are available in the medical field most of the data are not analysed for capturing valuable knowledges. Advanced techniques could be used to discover patterns and its relations. Our study shows  the implementation various of machine learning algorithms for developing a  breast cancer predictive model..

Method: The primary objectiveis to implement various ML techniques  such as Support Vector Machine (SVM) , Random Forest (RF) and Decision Tree(DT) to develop the predictive model and  to compare the parameters such as accuracy and performance of the different  algorithms.

Result and Conclusions: The results of our analysis indicate the accuracy of  SVM, RF and DT as 0.98, 0.9813 and 0.95 respectively.Based on the our results we concluded that  RF classification model  predicts with highest accuracy and least error rate in comparison with the other algorithms for breast cancer. It was also found  that DT showcased the least acccurate prediction  model for breast cancer.