الفهرس | Only 14 pages are availabe for public view |
Abstract Hepatocellular carcinoma (HCC) is a threat to the liver, which is considered one of the diseases devastating to human health that leads to death. HCC is a liver cancer that is considered one of the commonest causes of mortality worldwide. As a result, early detection of HCC is critical; this necessitates a mix of existing technique improvements and the creation of a novel, optimum approach. This will not begin without complete, adequate, and reliable data. Hence, it is imperative to improve missing data (MD) completion processes to provide more reliable data in the detection phase. The concept of MD is considered significant when applying statistical methods to a dataset and the quality of the data analysis results is based on the correct data completeness. As a result, improving MD filling processes is vital in order to provide more reliable data throughout the phase of analysis. Here, a novel method was presented for optimizing multiple regression imputation processes and obtaining the best fitness values for MD from patients by combining multiple imputations with a genetic algorithm (GA). To train and assess the proposed method, this work employed 583 patient records from a publicly available database, divided into 416 records of liver patients and 167 records of non-liver patients. The proposed approach offers the largest improvement for MD findings, according to the results. Instead of employing the normal equation in multiple imputations, which yielded 92.72 as the utmost fitness value with Mean Absolute Error (MAE) 0.5877 from 1.1840 after the second optimization, it enabled to achieve a fitness value of 233. An improved Multiple Imputation (MI) approach has been developed for obtaining the optimum fitness values while simultaneously enhancing Missing-Data (MD) completion procedures and giving more accurate data during the detection phase. To enhance multiple regression imputation procedures, multiple IV imputation has been integrated with a genetic approach. After solving MD phase with new optimized MI that resulted in the classification accuracy in positive direction and achieve 0.74 accuracy in the classification phase using TensorFlow DNN. Concluded that using the optimized MI with medical datasets for filling MD was the best option, which in turn affected to enhance the classification results in DNN. The proposed approach might be tested using a large database and used in HCC labs to help clinicians make accurate diagnoses. |