Construction of the prediction model for multiple myeloma based on machine learning

Area of Science:

• Hematology
• Machine Learning
• Medical Diagnostics

Background:

• Multiple myeloma (MM) poses a growing global health challenge.
• Early detection is crucial for improving patient outcomes.
• This study aimed to develop predictive models for MM early detection.

Purpose of the Study:

• To develop and validate machine learning models for early multiple myeloma detection.
• To compare the performance of logistic regression, support vector machine, and random forest models.
• To identify key hematological parameters for MM prediction.

Main Methods:

• Retrospective study involving 465 MM patients and 150 healthy controls.
• Developed logistic regression (LR), support vector machine (SVM), and random forest (RF) models using complete blood count (CBC) and cell population data (CPD).
• Validated models using receiver operating characteristic (ROC) curves, calibration curves, and decision curve analysis (DCA).

Main Results:

• Six parameters (RBC, RDW-CV, IG, NE-WZ, LY-WX, LY-WZ) were selected using LASSO.
• The random forest (RF) model achieved the highest Area Under the Curve (AUC) across training (0.956), validation (0.892), and test (0.875) sets.
• RF model demonstrated superior predictive performance compared to LR and SVM models.

Conclusions:

• The developed random forest model shows promise as an auxiliary tool for rapid multiple myeloma screening.
• Machine learning approaches can enhance early detection of hematological malignancies.
• Further validation in diverse clinical settings is warranted.