Calculating the probability of tumor control with machine learning methods

It is not possible to manually process and analyze large amounts of data. However, it is often necessary to make predictions for the future by using the past data. Machine learning methods are used to make these inferences. The systems where machine learning is used are decision support systems. Decision support systems vary depending on the field of use. Clinical decision support systems are used in medical applications. Conventional data analysis methods are ineffective against the complexity of medical data. For this reason, machine learning methods are used to understand and solve complex structures. The healthcare field is one of the areas where machine learning methods are used. Small cell lung cancer radiotherapy data were used in this study. Radiotherapy is also one of the areas in which many of the mentioned complex data are involved. A separate plan have been drawn for each patient who will receive radiotherapy. The response of the tumor to radiotherapy is different for each patien t. This may be due to the many parameters that affect radiotherapy, and the fact that each patient is different from one another in terms of their personal characteristics. Analyzing and examining complex data, machine learning methods are used to derive radiotherapy response of the tumor as a result of radiotherapy. In the study, it was aimed to classify and deduce patient data in radiation oncology by machine learning methods. Within the scope of the study, the data of the patients coming to the Radiation Oncology of the Karadeniz Technical University Faculty of Medicine Farabi Hospital between 2012-2015 had been used. In this context, classification and comparison of responses given by tumor supportive vector machines (SVM) and artificial neural networks (ANN) are discussed. In this study, two different decision support system models have been developed by means of SVM, ANN machine learning methods by using datas of 30 patient. While obtaining 90% sensivity, 100% specificity values for SVM model, for ANN model 80 % sensivity, 100 % specificity values have been obtained. In this study it have been seen that SVM model gave most succesful result with 90% sensivity. With this method, a prediction will be made about how much the patient who will receive radiotherapy will change in the tumor at the end of the treatment. In this study, the data were first subjected to preprocessing and processing. As a result, the number of data has decreased. New parameters should be added and the number of data should be 4 increased by considering these deficiencies for the success of the studies and the datas that completely should be obtained.