EPJ Web of Conferences 370, 01020 (2026)
https://doi.org/10.1051/epjconf/202637001020

Modeling of Non-Specific Liver Biomarkers for the Early Screening of Ovarian Cancer

Institute of Engineering and Management, University of Engineering and Management, Kolkata.

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Published online: 29 May 2026

Abstract

Ovarian Cancer (OC) cases are exponentially growing across the world and the mortality rate are also very alarming. The present screening and detection methods of OC cases includes a series of physical tests, imaging test followed by biopsy. Such tests are time consuming, costly and painful. This contributory study attempt to identify and modeling of various non-specific Liver bio-markers and its subsequent impacts on the early detection of the OC risk factors. Blood composed of various biomolecules such as plasma, proteins, osmotic pressure, hormones, enzymes, and antibodies. A secondary blood dataset was preprocessed and analyzed in this study followed by extraction of the key significant predictors for the modeling of various non-specific blood markers for the early screening of OC cases. Albumin, Alkaline Phosphate, Direct Bilirubin and Globulin were found as significant predictors biomolecules from the dataset for OC cases detection using rankers and best first attribute selection method. The study subsequently framed three machine learning models named Instance- Based k-nearest neighbors (IBK), Random Forest (RF) and Logistic Regression (LR) classifiers with each 10-fold cross validation and compares the model performances for the selection of the best framework for the early detection of OC cases. The study evident that the accuracy of the IBK, Random Forest and Logistic Regression frameworks are 59.8%, 69.6% and 73.3% respectively with ROC value0.60, 0.75 and 0.77 respectively. That evident the performance of LR is best among other models for the accurately detection of the cases of OC with high true positive rate (76.4%) and low false positive rate. The study shows a pathway on combining features from various non-specific biomarkers for screening the risk factor of OC cases at early stage. The authors are hopeful such low cost, less invasive approaches with promising pattern finding methods of machine learning, will appreciate by medical professionals through clinical studies and tests that will help to detect OC cases at very beginning for better treatment outcome.