A Comprehensive Reactive-Oxygen-Species-Scavenging Metal-Organic Framework Orchestrates the AMPK-DOT1L-H3K79me3 Cascade to Alleviate Osteoarthritis.

Tuberculosis (TB) is one of the oldest known diseases, affecting humans since prehistoric times, as evidenced by characteristic lesions found in Neolithic skeletal remains [, ]. Spinal tuberculosis, specifically known as Pott’s disease, accounts for approximately 50% of all musculoskeletal TB cases [–]. It is estimated that approximately one-quarter of the global population has been infected with , extrapulmonary manifestations occur in 20% of cases, and skeletal involvement is seen in nearly 10% of active patients. The spine is the most common site of skeletal involvement [–]. The onset of spinal tuberculosis is slow and insidious. Patients often present with atypical back pain. Kyphosis and severe neurological complications may occur if not diagnosed and treated in time. Paraplegia is the most serious complication after nervous system damage, with an incidence of 10–30% [, –]. The thoracic spine (40–50%) and lumbar spine (35–45%) are common sites of spinal involvement []. In clinical practice, for patients with spinal tuberculosis who have clear surgical indications, standardized anti-tuberculosis drug treatment should be performed first, and then surgical treatment, including abscess drainage, lesion debridement, bone graft fusion, and internal fixation placement, etc. Thus, it can improve the efficacy of anti-tuberculosis drugs, strengthen the control of tuberculosis, promote bone graft fusion, and reconstruct the stability of the spine [–]. TB bacteria mainly attack people with poor basic conditions, such as patients with immunosuppression, malnutrition, organ dysfunction, diabetes, or a history of smoking and alcohol abuse [, ]. Therefore, such patients are more prone to various postoperative complications compared to patients with spinal degenerative diseases [, ]. Poor healing of wound (PWH) is a common but serious postoperative complication that serves as a major risk factor for surgical site infection (SSI). PWH can lead to internal fixation failure, pseudarthrosis, and osteomyelitis, often necessitating secondary revision surgery. Consequently, this prolongs hospital stays, increases economic burdens, and elevates mortality rates [–]. Controlling systemic toxicity and optimizing nutritional status are pivotal in the management of spinal tuberculosis. Previous studies have demonstrated that nutritional status significantly influences postoperative wound healing []. Following effective control of tubercular toxicity, preoperative nutritional assessment becomes critical. Therefore, patients should undergo a comprehensive preoperative nutritional status assessment, and nutritional, immune-inflammatory and surgery-related markers should be used to evaluate the possibility of poor postoperative wound prognosis. The Nutritional Risk Screening 2002 (NRS-2002), Patient-Generated Subjective Global assessment (PG-SGA) and other assessment methods have been widely used in clinical practice. However, these methods are still limited because they include multiple subjective factors and require expertise for accurate assessment. The Prognostic Nutritional Index (PNI) based on serum albumin and peripheral blood lymphocyte count was originally used to assess preoperative nutritional status, surgical risk, and postoperative complications in surgical patients. It has been shown to be a prognostic biomarker for postoperative spinal infection, postoperative delirium, solid tumors, and cardiovascular disease []. The Naples Prognostic Score (NPS), first proposed by Gennaro et al. [], has been identified as an independent prognostic factor for patients undergoing colorectal cancer surgery. It has demonstrated good prognostic value in gastrointestinal tumors []. However, the predictive value of PNI and NPS for wound healing outcomes following posterior instrumentation and debridement for thoracolumbar tuberculosis remains unclear. Currently, while studies have explored risk factors associated with postoperative wound healing, there is a lack of comprehensive, individualized evaluation methods for accurate prediction. Postoperative wound healing is influenced by complex, confounding risk factors that are difficult to predict preoperatively using traditional methods alone. Standard logistic regression analysis, while common, may fail to fully capture the intricate interactions between risk factors. In contrast, machine learning (ML) techniques offer significant advantages over traditional statistical methods. ML algorithms utilize powerful computational capabilities to automatically detect complex, nonlinear relationships between features without the strict assumptions required by conventional statistics. This allows ML models to handle high-dimensional data and multicollinearity more effectively. For example, deep learning models can be used to extract abstract feature representations of the data through multi-level nonlinear transformations, enabling more efficient processing of highly correlated features. Therefore, we i