APPLICATION OF ARTIFICIAL INTELLIGENCE AND MACHINE LEARNING IN THE PREDICTION AND MANAGEMENT OF COMPLICATIONS AND MORBIDITY AND MORTALITY IN POLYTRAUMATIZED PATIENTS: A SYSTEMATIC LITERATURE REVIEW

• DOI: https://doi.org/10.22533/at.ed.1595142512035

• Palavras-chave: Artificial Intelligence. Machine Learning. Multiple Trauma. Morbidity and mortality indicators.

• Keywords: Artificial Intelligence. Machine Learning. Multiple Trauma. Morbidity and mortality indicators.

• Abstract:

  INTRODUCTION: Artificial Intelligence (AI) and Machine Learning (ML) are emerging tools in medicine, capable of analyzing large volumes of clinical data to identify patterns and predict patient outcomes. In trauma medicine, where the management of polytraumatized patients is particularly complex, accurate prediction of complications is essential. Although the Trauma and Injury Severity Score (TRISS) is widely used as a predictive method, it has important limitations, such as its inability to capture the dynamic and individual complexity of trauma responses. OBJECTIVES: To evaluate the effectiveness of AI and ML-based tools in predicting complications in polytraumatized patients. METHODS: The following descriptors were searched in the Cochrane, Scopus and PubMed databases: "Artificial Intelligence", "Machine Learning", "polytrauma" and "predictive analytics", using the "AND" operator. Studies addressing the use of AI and ML in the management of polytraumatized patients were included, covering systematic reviews, clinical trials, meta-analyses and longitudinal observational studies published between 2019 and 2024. Duplicate, unrelated and unavailable studies were excluded. Data extraction was carried out by two independent authors. RESULTS: Of the 18 studies identified, 14 were included in the review. DISCUSSION: Studies such as Gondim et al. (2023) have shown that ML algorithms outperform traditional models such as TRISS in predicting mortality in real time. Unlike linear models, ML considers risk factors in a non-linear and interactive way, better reflecting clinical reality. ML has also been effective in predicting specific complications, such as acute traumatic coagulopathy (Wang, 2020), venous thromboembolism (Li, 2021), acute kidney injury (Liu, 2022) and sepsis (Gelbard, 2019). AI-based tools have also proven useful in contexts of limited resources and war situations (Perkins, 2021), optimizing management and resource allocation. Innovations such as risk calculators for mobile devices (Maurer, 2021) have facilitated access to pre-hospital care. CONCLUSION: Artificial Intelligence and Machine Learning show great potential in predicting complications in polytraumatized patients, offering greater accuracy compared to traditional models.