Machine Learning-based Prediction Model for Treatment of Acromegaly With First-generation Somatostatin Receptor...

Area of Science:

• Endocrinology
• Medical Artificial Intelligence
• Biomarker Analysis

Background:

• Acromegaly treatment response to first-generation somatostatin receptor ligands (fg-SRLs) can vary.
• Predicting therapeutic outcomes is crucial for optimizing patient management.

Purpose of the Study:

• Develop a machine learning (ML) model to predict treatment response in acromegaly patients receiving fg-SRLs.
• Identify key biomarkers for predicting therapeutic success.

Main Methods:

• Evaluated six ML models (logistic regression, k-nearest neighbor, support vector machine, gradient-boosted, random forest, multilayer perceptron).
• Utilized patient data including age, sex, growth hormone (GH), insulin-like growth factor-I (IGF-I) levels, somatostatin receptor (SST2, SST5) expression, and cytokeratin granulation pattern (GP).

Main Results:

• A support vector machine model achieved 86.3% accuracy, predicting response based on SST2, SST5, GP, sex, age, and pretreatment GH/IGF-I levels.
• Controlled patients were older, had lower diagnostic/pretreatment GH and IGF-I, and exhibited densely granulated tumors with high SST2 expression.

Conclusions:

• Developed a highly accurate ML-based prediction model for acromegaly treatment response.
• The model has the potential to enhance acromegaly management, optimize biochemical control, and reduce long-term complications and healthcare costs.