# Development and validation of prognostic prediction model for submandibular gland cancer based on the SEER database

## TL;DR

This study developed a reliable model to predict the prognosis of submandibular gland cancer patients using data from the SEER database.

## Contribution

A novel prognostic prediction model for submandibular gland cancer using lasso and Cox regression was developed and validated.

## Key findings

- The model identified age, tumor size, histology, and lymph node metastasis as key prognostic factors.
- The model showed strong discrimination with c-indexes of 0.802 in training and 0.756 in validation sets.
- Decision Curve Analysis confirmed the model's clinical utility for guiding treatment decisions.

## Abstract

•Prognosis of submandibular gland carcinoma remains challenging to predict.•Lasso regression selected key prognostic factors.•Cox regression model established for prognostic prediction.•The model demonstrated accuracy with c-indexes in training and validation sets.•Decision curve analysis confirmed the model’s clinical utility.

Prognosis of submandibular gland carcinoma remains challenging to predict.

Lasso regression selected key prognostic factors.

Cox regression model established for prognostic prediction.

The model demonstrated accuracy with c-indexes in training and validation sets.

Decision curve analysis confirmed the model’s clinical utility.

Accurately predicting the prognosis of Submandibular Gland Carcinoma (SGC) patients remains a challenging task. The purpose of this study was to develop a columnar graph prognostic prediction model for submandibular gland cancer based on the SEER database, using feature selection with lasso regression and modeling with Cox regression.

This study utilized data from the SEER database, focusing on 1362 cases of SGC. Various clinical and demographic factors, including age, tumor size, histology, and lymph node metastasis, were considered as potential prognostic factors. Feature selection was performed using lasso regression, and a Cox proportional hazards model was constructed, taking into account the complex interactions between variables and their impact on survival outcomes.

The established prognostic prediction model demonstrated good accuracy and reliability. The model effectively identified several important prognostic factors, including age, tumor size, histology, and lymph node metastasis, which strongly influenced the prognosis of SGC. The model showed good discrimination and calibration with c-indexes of 0.802 (0.784‒0.821) in the training set and 0.756 (0.725‒0.787) in the validation set. The Decision Curve Analysis (DCA) curve reflected clinical utility.

This study suggests that the prognostic prediction model based on Cox regression is a valuable tool for predicting the prognosis of patients with SGC. This approach has the potential to improve patient outcomes by facilitating personalized treatment plans and identifying high-risk patients who may benefit from more aggressive interventions.

Level III.