Missing Data Estimation for Cancer Diagnosis Support

The paper presents a heterogeneous neural network based system that can be used for estimation of missing tumor marker values in patient data and in a second step for calculating the possibility of a cancerous disease. For estimation of missing values we use different approaches: neural network based estimation of a specific marker depending on existing values of a related marker and neural network based estimation of missing tumor markers depending on standard blood parameter measurements. Finally we compare the results for calculating the possibility of a cancerous disease using different methods for missing value estimation of patient data.