Predicting one-year overall survival in patients with AITL using machine learning algorithms: a multicenter study

Area of Science:

• Hematology
• Oncology
• Computational Biology

Background:

• Angioimmunoblastic T-cell lymphoma (AITL) is a severe hematological malignancy.
• Patients with poor prognosis have limited treatment benefits from traditional therapies.
• Accurate survival prediction is crucial for guiding treatment decisions in AITL.

Purpose of the Study:

• To develop an interpretable machine learning (ML) model for predicting 1-year overall survival (OS) in AITL patients.
• To identify key baseline characteristics influencing AITL patient survival.
• To enhance prognostic accuracy for AITL management.

Main Methods:

• Utilized data from 223 AITL patients across 4 Chinese centers.
• Developed and compared five ML algorithms for 1-year OS prediction.
• Employed Recursive Feature Elimination (RFE) for feature selection and SHAP/LIME for model interpretability.

Main Results:

• The Catboost model achieved the highest predictive performance with an AUC of 0.8277.
• After RFE screening, a model with 8 key variables demonstrated strong predictive power (AUC = 0.8125).
• Interpretable ML methods confirmed the relevance of selected prognostic factors.

Conclusions:

• An interpretable Catboost model incorporating 8 variables effectively predicts 1-year OS in AITL patients.
• This ML-based approach can aid clinicians in risk stratification and treatment planning.
• The study highlights the potential of AI in improving outcomes for hematological malignancies.