Using artificial intelligence and statistics for managing peritoneal metastases from gastrointestinal cancers

Area of Science:

• Oncology
• Medical Informatics
• Computational Biology

Background:

• Peritoneal metastases (PM) are a significant challenge in gastrointestinal (GI) cancers, impacting patient prognosis and treatment strategies.
• Accurate detection and management of PM are crucial for improving clinical decision-making and patient outcomes.
• Traditional statistical methods have limitations in analyzing complex datasets associated with PM.

Purpose of the Study:

• To investigate the application of artificial intelligence (AI) and statistical methodologies for analyzing and managing peritoneal metastases (PM) in gastrointestinal cancers.
• To compare the efficacy of conventional machine learning (ML) and deep learning (DL) models against traditional statistical approaches for PM detection.
• To identify factors influencing the predictive accuracy of AI and statistical models in the context of PM.

Main Methods:

• A systematic literature review was conducted using PubMed and Google Scholar with predefined keywords and search criteria.
• Studies incorporating AI (machine learning and deep learning) and statistical (biostatistics, logistic models) approaches for PM analysis were included.
• Performance metrics and influencing factors, such as sample size, were analyzed across the selected studies.

Main Results:

• AI methodologies, including deep learning (DL) and machine learning (ML), demonstrated superior performance in detecting and analyzing peritoneal metastases compared to traditional statistical methods.
• Deep learning models consistently yielded the most precise results, while classical ML models showed high predictive accuracy with varied performance.
• Larger sample sizes were identified as a key factor in enhancing the predictive accuracy of both AI and statistical models.

Conclusions:

• AI and statistical approaches are effective tools for detecting peritoneal metastases in gastrointestinal cancers, aiding in diagnostics and prognostics.
• Integration of AI into clinical practice can enhance the analysis and management of PM, leading to improved clinical decision-making and patient outcomes.
• Multicenter collaboration is recommended to standardize data collection methods, ensuring consistent and reliable results for AI and statistical model development.