The predictive power of artificial intelligence on mediastinal lymphnode metastasis

Area of Science:

• Oncology
• Radiology
• Artificial Intelligence

Background:

• Mediastinal lymph node metastasis is crucial in lung adenocarcinoma staging.
• Accurate preoperative prediction aids treatment planning.
• Current methods have limitations in predicting metastasis.

Purpose of the Study:

• To develop an artificial intelligence (AI)-based preoperative predictive model for mediastinal lymph node metastasis in lung adenocarcinoma.
• To evaluate the AI model's performance against traditional methods.

Main Methods:

• A cohort of 301 patients with resected lung adenocarcinoma (clinical stage N0-1) was analyzed.
• Patients were divided into training (n=201) and validation (n=100) sets.
• An automatic machine learning platform was used to create the AI model.
• Multivariate analysis identified predictive factors, including maximum standardized uptake value (SUVmax).

Main Results:

• The AI model achieved 84% accuracy and 98% specificity.
• SUVmax showed 61% accuracy and 57% specificity.
• The AI model's sensitivity was notably low at 12%.

Conclusions:

• AI demonstrates high specificity in predicting mediastinal lymph node metastasis in lung adenocarcinoma.
• The AI model shows potential to enhance existing diagnostic tools.
• Further research is needed to improve sensitivity for clinical readiness.