Machine learning for predicting overall survival in early-stage supraglottic cancer: a SEER-based population study

Area of Science:

• Oncology
• Machine Learning in Medicine
• Cancer Prognostics

Background:

• Supraglottic squamous cell carcinoma (SGSCC) is the second most common laryngeal cancer.
• SGSCC is associated with a poor prognosis, necessitating improved predictive models.

Purpose of the Study:

• To develop a machine learning (ML) model integrating clinicopathological and treatment factors for overall survival (OS) prediction in early-stage SGSCC.
• To identify key prognostic factors influencing OS in this patient cohort.

Main Methods:

• Utilized data from 1171 SGSCC patients from the SEER database.
• Employed feature importance analysis to identify significant predictors of OS within the ML model.

Main Results:

• The ML model achieved a weighted accuracy of 72.3% in predicting OS.
• Top predictors for OS included age at diagnosis, marital status, number of malignancies, regional lymph nodes, and radiotherapy.
• Older age correlated with decreased OS, while marriage, absence of other malignancies, surgery, and radiotherapy were linked to improved survival.

Conclusions:

• Integrating clinicopathological and treatment data effectively predicts OS in early-stage SGSCC.
• Further external validation is recommended to confirm the generalizability of the developed ML model.