Abstract
Background: Machine Learning (ML) is rapidly growing in capability and is increasingly applied to model outcomes and complications in medicine. Surgical site infections (SSI) are a common postoperative complication in spinal surgery. This study aimed to develop and validate supervised ML algorithms for predicting the risk of SSI following minimally invasive transforaminal lumbar interbody fusion (MIS-TLIF) under the Quadrant channel.
Methods: This single-central retrospective study included a total of 705 cases between May 2012 and October 2019. Data of patients who underwent MIS-TLIF under the Quadrant channel was extracted by the electronic medical record system. The patient’s clinical characteristics, surgery-related parameters, and routine laboratory tests were collected. Univariate and multivariate logistic regression analyses were used to screen and identify independent risk factors for SSI. Then, the independent risk factors were imported into six ML algorithms, including k-Nearest Neighbor (KNN), Decision Tree (DT), Support Vector Machine (SVM), Random Forest (RF), Multi-Layer Perceptron (MLP), and Naïve Bayes (NB), to develop a prediction model for predicting the risk of SSI following MIS-TLIF under Quadrant channel. During the training process, 10-fold cross-validation was used for validation. Indices like the area under the receiver operating characteristic (AUC), sensitivity, specificity, and accuracy (ACC) were reported to test the performance of ML models.
Results: Among the 705 patients, SSI occurred in 33 patients (4.68%). The univariate and multivariate logistic regression analyses showed that preoperative glycated hemoglobin A1c (HbA1c), estimated blood loss (EBL), preoperative albumin, body mass index (BMI), and age were all independent predictors of SSI. In predicting SSI, six ML models posted an average AUC of 0.60-0.80 and an ACC of 0.80-0.95, with the NB model standing out, registering an average AUC and an ACC of 0.78 and 0.90. Then, the feature importance of the NB model was reported.
Conclusions: ML algorithms are impressive tools in clinical decision-making, which can achieve satisfactory prediction of SSI with the NB model performing the best. The NB model may help access the risk of SSI following MIS-TLIF under the Quadrant channel and facilitate clinical decision-making. However, future external validation is needed.
Introduction
Surgical site infections (SSI) represent a significant burden in spinal surgery, often leading to prolonged hospital stays, increased healthcare costs, and higher patient mortality rates. While Minimally Invasive Transforaminal Lumbar Interbody Fusion (MIS-TLIF) is specifically designed to reduce tissue trauma compared to traditional open surgery, infections still occur in approximately 2% to 5% of cases.
Existing medical literature often identifies individual risk factors such as obesity or diabetes using simple odds ratios. However, these isolated factors do not provide a cohesive, individualized risk profile for patients. This study sought to bridge this gap by developing a predictive tool specifically for MIS-TLIF performed under a "Quadrant channel" (a specialized surgical access tool), providing clinicians with a data-driven method to personalize patient care.
Methods and Study Population
The researchers analyzed a retrospective dataset from Taizhou Central Hospital, covering 705 patients who underwent MIS-TLIF between May 2012 and October 2019. Of these, 33 patients (4.68%) developed an SSI. The study examined a wide range of features, including clinical characteristics (age, BMI), surgery-related parameters (operation time, blood loss), and laboratory results (HbA1c, albumin, fibrinogen).
The researchers employed a two-stage approach. First, traditional statistical methods—univariate and multivariate logistic regression—were used to identify the most significant independent predictors. Second, these predictors were fed into six supervised machine learning (ML) algorithms: k-Nearest Neighbor (KNN), Decision Tree (DT), Support Vector Machine (SVM), Random Forest (RF), Multi-Layer Perceptron (MLP), and Naïve Bayes (NB). To prevent overfitting, the models were validated using 10-fold cross-validation.
Key Results and Predictors
The study identified several critical variables that significantly influenced the risk of infection. Notably, metabolic health and surgical efficiency emerged as primary drivers of SSI risk.
As shown in the data, the most influential variables for predicting SSI were preoperative HbA1c, Estimated Blood Loss (EBL), and Operation Time. The Naïve Bayes (NB) model emerged as the most effective algorithm, achieving an Area Under the Curve (AUC) of 0.78 and an overall accuracy of 0.90. Multivariate analysis confirmed that advanced age (OR 1.187), high BMI (OR 1.445), and low preoperative albumin (OR 0.672) significantly increased the odds of infection.
Discussion
Metabolic Health is Paramount
Preoperative HbA1c, a measure of long-term blood sugar control, was the single most important predictor. This suggests that even in minimally invasive procedures, a patient's underlying metabolic state is a primary determinant of their immune system's ability to prevent infection. Optimizing blood sugar levels before surgery may be a key intervention for high-risk patients.
The Power of Simplicity
Interestingly, the Naïve Bayes model outperformed more complex "black box" algorithms like Random Forests or Neural Networks. In clinical settings with relatively small datasets—particularly when the "positive" outcome (infection) is rare—simpler probabilistic models often prove more robust and less prone to noise.
Limitations and Future Work
While promising, the study has limitations. As a retrospective, single-center study, it may contain selection biases. Furthermore, the relatively low number of infection cases (33) means the "signal" for infection is subtle. Future research involving larger, multi-center cohorts is necessary to externally validate these findings and improve the model's sensitivity across diverse patient populations.
Conclusion
Machine learning algorithms, particularly the Naïve Bayes model, are valuable tools for clinical decision-making in spinal surgery. By identifying high-risk patients before they enter the operating room, surgeons can implement targeted preventive measures to reduce the incidence of SSI following MIS-TLIF under the Quadrant channel.
Summary updated