Related Experiment Video
Updated: Jun 21, 2025
Implantation and Evaluation of Melanoma in the Murine Choroid via Optical Coherence Tomography
Published on: December 2, 2022
Uveal melanoma distant metastasis prediction system: A retrospective observational study based on machine learning
Shi-Nan Wu1, Dan-Yi Qin1, Linfangzi Zhu1
1Xiamen University Affiliated Xiamen Eye Center, Fujian Provincial Key Laboratory of Ophthalmology and Visual Science, Fujian Engineering and Research Center of Eye Regenerative Medicine, Eye Institute of Xiamen University, School of Medicine, Xiamen University, Xiamen, Fujian, China.
Machine learning accurately predicts distant metastasis in uveal melanoma (UM) patients. A multilayer perceptron model offers personalized risk assessment, aiding early treatment strategies for better prognosis.
Area of Science:
- Oncology
- Data Science
- Medical Informatics
Background:
- Uveal melanoma (UM) carries a high risk of distant metastasis, leading to poor patient prognosis.
- Predictive models for UM distant metastasis are lacking, especially those utilizing big data.
- Early identification of metastasis risk is crucial for timely intervention.
Purpose of the Study:
- To develop and validate a machine learning model for predicting distant metastasis in UM patients.
- To identify key risk and protective factors associated with UM distant metastasis.
- To create a personalized risk assessment tool for UM patients.
Main Methods:
- Utilized the Surveillance, Epidemiology, and End Results (SEER) database (2000-2020) for UM patient data.
- Applied logistic regression and six machine learning algorithms, selecting the multilayer perceptron (MLP) model.
- Employed Shapley additive explanations (SHAP) for model interpretation and developed a web-based calculator.
Main Results:
- The MLP model achieved high predictive accuracy (ROC AUC = 0.876).
- Key risk factors identified include grade, age, primary site, time to treatment, and tumor count.
- Protective factors include diagnostic method, laterality, rural-urban code, and radiation therapy.
Conclusions:
- The MLP model is an optimal tool for predicting distant metastasis in UM.
- Personalized risk assessments can be achieved through the developed web calculator.
- This facilitates earlier and tailored treatment strategies for improved patient outcomes.
