Non-Invasive Cancer Detection Using Blood Test and Predictive Modeling Approach.

Area of Science:

• Oncology
• Bioinformatics
• Machine Learning

Background:

• Rising cancer incidence presents a significant public health challenge.
• Early and accurate diagnosis is crucial for effective cancer treatment and patient outcomes.

Purpose of the Study:

• To develop a machine learning-driven predictive model for simultaneous diagnosis of multiple cancer types.
• To integrate hematological parameters with machine learning for non-invasive cancer detection.

Main Methods:

• Analysis of a dataset of 19,537 laboratory reports from Jordanian hospitals.
• Data preprocessing including feature standardization and missing value imputation.
• Application of machine learning classifiers such as Random Forest, Linear Discriminant Analysis, Support Vector Machine, and Histogram Gradient Boosting.

Main Results:

• Hematological features like white blood cell count, red blood cell count, and platelet count, along with age and creatinine, were key predictors.
• Random Forest, LDA, and SVM achieved high prediction accuracy (0.69-0.72).
• The Histogram Gradient Boosting model demonstrated improved performance.

Conclusions:

• The integration of hematological indicators and machine learning provides an efficient platform for non-invasive cancer screening.
• Future research exploring deep learning could enhance prediction accuracy by identifying complex patterns.