Artificial Intelligence Applied in Early Prediction of Lower Limb Fracture Complications.

Abstract

Background: Artificial intelligence has become a valuable tool for diagnosing and detecting postoperative complications early. Through imaging and biochemical markers, clinicians can anticipate the clinical progression of patients and the risk of long-term complications that could impact the quality of life or even be life-threatening. In this context, artificial intelligence is crucial for identifying early signs of complications and enabling clinicians to take preventive measures before problems worsen. Materials and methods: This observational study analyzed medical charts from the electronic archive of the Clinical Emergency Hospital in Galați, Romania, covering a four-year period from 2018 to 2022. A neural network model was developed to analyze various socio-demographic and paraclinical data. Key features included patient demographics, laboratory investigations, and clinical outcomes. Statistical analyses were performed to identify significant risk factors associated with deep venous thrombosis (DVT). Results: The analysis revealed a higher prevalence of female patients (60.78%) compared to male patients, indicating a potential gender-related risk factor for DVT. The incidence of DVT was highest among patients aged 71 to 90 years, affecting 56.86% of individuals in this age group, suggesting that advanced age significantly contributes to the risk of developing DVT. Additionally, among the DVT patients, 15.69% had a body mass index (BMI) greater than 30, categorizing them as obese, which is known to increase the risk of thrombotic events. Furthermore, this study highlighted that the highest frequency of DVT was associated with femur fractures, occurring in 52% of patients with this type of injury. The neural network analysis indicated that elevated levels of direct bilirubin (≥1.5 mg/dL) and prothrombin activity (≤60%) were strong predictors of fracture-related complications, with sensitivity and specificity rates of 78% and 82%, respectively. These findings underscore the importance of monitoring these laboratory markers in at-risk populations for early intervention. Conclusions: This study identified critical risk factors for developing DVT, including advanced age, high BMI, and femur fractures, which necessitate longer recovery periods. Additionally, the findings indicate that elevated direct bilirubin and prothrombin activity play a significant role in predicting DVT development. These results suggest that AI can effectively enhance the anticipation of clinical evolution in patients, aiding in early intervention and management strategies.