An Artificial Intelligence-Based Diagnostic System for Acute Lymphoblastic Leukemia Detection.

Area of Science:

• Hematology
• Medical Informatics
• Machine Learning in Medicine

Background:

• Acute Lymphoblastic Leukemia (ALL) is a significant hematologic malignancy.
• Accurate and early diagnosis of ALL is crucial for effective treatment.
• Current diagnostic methods may be invasive or costly.

Purpose of the Study:

• To develop a novel diagnostic model for ALL using only complete blood count (CBC) data.
• To identify key CBC parameters specific to ALL diagnosis.
• To compare the efficacy of different machine learning algorithms for ALL detection based on CBC data.

Main Methods:

• A dataset of 86 ALL patients and 86 controls was utilized.
• Feature selection techniques were employed to identify ALL-specific CBC parameters.
• Random Forest, XGBoost, and Decision Tree algorithms were trained and tuned using Grid Search and 5-fold cross-validation.
• Model performance was evaluated for ALL detection.

Main Results:

• Specific CBC parameters were identified as highly indicative of ALL.
• The Decision Tree classifier achieved superior performance in ALL detection compared to XGBoost and Random Forest.
• The developed model demonstrates the potential of CBC data for non-invasive ALL diagnosis.

Conclusions:

• A novel and effective diagnostic model for ALL can be constructed using solely CBC data.
• The Decision Tree algorithm shows promise for accurate ALL detection based on CBC profiles.
• This approach offers a potentially simpler and more accessible method for ALL screening and diagnosis.