The application of random forest-based models in prognostication of gastrointestinal tract malignancies: a systematic review

Abstract

Introduction: Malignancies of the GI tract account for one-third of cancer-related deaths globally and more than 25% of all cancer diagnoses. The rising prevalence of GI tract malignancies and the shortcomings of existing treatment approaches highlight the need for better predictive prediction models. RF's machine-learning method can predict cancers by using numerous decision trees to locate, classify, and forecast data. This systematic study aims to assess how well RF models predict the prognosis of GI tract malignancies.

Methods: Following PRISMA criteria, we performed a systematic search in PubMed, Scopus, Google Scholar, and Web of Science until May 28, 2024. Studies used RF models to forecast the prognosis of GI tract malignancies, including esophageal, gastric, and colorectal cancers. The QUIPS approach was used to evaluate the quality of the included studies.

Results: Out of 1846 records, 86 studies met inclusion requirements; eight were disqualified. Numerous studies showed that when combining clinical, genetic, and pathological data, RF models were very accurate and dependable in predicting the prognosis of GI tract malignancies, responses, recurrence, survival rates, and metastatic risks, distinguishing between operable and inoperable tumors, and patient outcomes. RF models outperformed conventional prognostic techniques in terms of accuracy; several research studies reported prediction accuracies of over 80% in survival rate estimates.

Conclusion: RF models, in terms of accuracy, performed better than the conventional approaches and provided better capabilities for clinical decision-making. Such models can increase the life quality and survival of patients by personalizing their treatment regimens for cancers of the GI tract. These models can, in a significant manner, raise patients' survival and quality of life through hastening clinical decision-making and providing personalized treatment options.

Keywords: GI tract cancers; malignancy; prognose; prognostication; random forest.

Figure 1

PRISMA 2020 flow diagram for new systematic reviews which included searches of databases.

Figure 2

Schematic illustration of an RF model: combining multiple decision trees to improve accuracy and reduce overfitting in predictions.