Multilevel Fusion Research Articles

Subcutaneous Lumbar Spine Index (SLSI) as a Risk Factor for Surgical Site Infection After Lumbar Fusion Surgery: A Retrospective Matched Case-Control Study.

A retrospective matched case-control study. This study aims to investigate the value of Subcutaneous Lumbar Spine Index (SLSI) as a predictor of early surgical site infection (SSI) after lumbar intervertebral fusion surgery. A retrospective case-control study was performed on patients who underwent transforaminal lumbar interbody fusion (TLIF) from January 1, 2014 to December 31, 2019 in a single institution. Cases were defined as those who developed early SSI according to the US Center for Disease Control and Prevention criteria, and controls were matched from those patients without early SSI using the following matched criteria: gender, age, time of surgery and diabetes. Subcutaneous fat thickness (SFT) and SLSI were measured on preoperative MRI mid-sagittal T2 weighted images. A total of 3615 patients who underwent TLIF were enrolled in this study. Thirty-three patients were included in early SSI, and sixty-six patients were selected as matched controls. Univariate analysis indicated that fusion levels (P = .007), operation time (P = .022), obesity (P = .013), SFT (P = .002) and SLSI (P = .001) were significantly associated with early SSI. Multiple logistic regression analysis revealed that multilevel fusion levels (P = .021), obesity (P = .035), a large SFT (P = .026) and a high SLSI (P = .012) were independent risk factors. Body mass index (BMI) and SLSI were moderately correlated (r2 = .55). ROC curve demonstrated that SLSI was more sensitive than SFT to predict the early SSI. SLSI is a novel radiological risk factor for early SSI development and is a better indicator than SFT to predict early SSI risk after lumbar intervertebral fusion.

Risk factors for extended length of stay and non-home discharge in adults treated with multi-level fusion for lumbar degenerative pathology and deformity

PurposeTo identify independent risk factors, including the Risk Assessment and Prediction Tool (RAPT) score, associated with extended length of stay (eLOS) and non-home discharge following elective multi-level instrumented spine fusion operations for diagnosis of adult spinal deformity (ASD) and lumbar degenerative pathology.MethodsAdults who underwent multi-level (ge 3 segments) instrumented spine fusions for ASD and lumbar degenerative pathology at a single institution (2016–2021) were reviewed. Presence of a pre-operative RAPT score was used as an inclusion criterion. Excluded were patients who underwent non-elective operations, revisions, operations for trauma, malignancy, and/or infections. Outcomes were eLOS (> 7 days) and discharge location (home vs. non-home). Predictor variables included demographics, comorbidities, operative information, Surgical Invasiveness Index (SII), and RAPT score. Fisher’s exact test was used for univariate analysis, and significant variables were implemented in multivariate binary logistic regression, with generation of 95% percent confidence intervals (CI), odds ratios (OR), and p-values.ResultsIncluded for analysis were 355 patients. Post-operatively, 36.6% (n = 130) had eLOS and 53.2% (n = 189) had a non-home discharge. Risk factors significant for a non-home discharge were older age (> 70 years), SII > 36, pre-op RAPT < 10, DMII, diagnosis of depression or anxiety, and eLOS. Risk factors significant for an eLOS were SII > 20, RAPT < 6, and an ASA score of 3.ConclusionThe RAPT score and SII were most important significant predictors of eLOS and non-home discharges following multi-level instrumented fusions for lumbar spinal pathology and deformity. Preoperative optimization of the RAPT’s individual components may provide a useful strategy for decreasing LOS and modifying discharge disposition.

Online course recommendation algorithm based on multilevel fusion of user features and item features

AbstractTo meet the demand for accurate recommendation and personalized learning in online education, an online course recommendation algorithm with multilevel fusion of user features and item features is proposed for content‐based recommendation systems applied to online education courses that have weak generalization ability and cannot cope well with data sparsity. The algorithm is improved on the deep learning recommendation algorithm named Wide &amp; Deep Learning (WDL), which is a CTR (Click‐through rate prediction) prediction algorithm. MMF (Multi‐level Fusion Feature) uses collaborative filtering to replace linear methods in the wide part of WDL, introduces feature interactions to model user representations and item representations separately, and uses ResNet (Residual Network) ideas to improve deep neural network (DNN) in the deep part to reduce the performance degradation caused by overfitting. The experimental validation was conducted on the online education data set and the public data set MovieLens‐1M, and the AUC was improved by 1.21% and 1.46%, respectively. Meanwhile, the effect brought by this improved algorithm is interpretable.

SiamMMF: multi-modal multi-level fusion object tracking based on Siamese networks

SiamMMF: multi-modal multi-level fusion object tracking based on Siamese networks

M-LFM: a multi-level fusion modeling method for shape-performance integrated digital twin of complex structure

M-LFM: a multi-level fusion modeling method for shape-performance integrated digital twin of complex structure

Upper Instrumented Vertebrae, Number of Levels Fused, and Plate Morphology Do Not Affect Severity of Chronic Dysphagia After Anterior Cervical Spine Surgery

Disordered swallowing, or dysphagia, is the most common complication after anterior cervical spine (ACS) surgery. Many operative factors are associated with development of dysphagia. The aim of this study was to explore how number of levels instrumented, specific levels fused, and plate morphology affect chronic dysphagia after ACS surgery. Consecutive patients referred to a tertiary center for otolaryngology evaluation for chronic dysphagia after ACS surgery between 2012 and 2017 were enrolled. Basic demographic data were obtained. Plain radiographs were reviewed for number of levels fused, upper instrumented vertebrae, and plate morphology. Plate morphology was categorized as a closed, small window, large window, or no profile plate. The 10-item Eating Assessment Tool (EAT-10) was used to assess dysphagia severity. Of 171 patients referred for dysphagia, 126 met inclusion criteria; 54% were female, with a mean age of 63 years (range, 32-88 years). Mean EAT-10 score was 18.5 ± 10.1. Mean time from spine surgery to videofluoroscopic swallow study for dysphagia was 58.3 months. Mean number of levels fused was 2.2 ± 0.9. There were no significant differences in EAT-10 scores in single-level versus multilevel fusion (19.0 vs. 18.4, P= 0.76) as well as with regard to upper instrumented vertebrae or plate morphology. In this series of patients with chronic dysphagia following ACS surgery, the severity of dysphagia as measured by the EAT-10 was not affected by upper instrumented vertebrae, number of levels fused, or plate morphology.

Machine Learning-Based Multi-Level Fusion Framework for a Hybrid Voltammetric and Impedimetric Metal Ions Electronic Tongue

Electronic tongues and artificial gustation for crucial analytes in the environment, such as metal ions, are becoming increasingly important. In this contribution, we propose a multi-level fusion framework for a hybrid impedimetric and voltammetric electronic tongue to enhance the accuracy of K+, Mg2+, and Ca2+ detection in an extensive concentration range (100.0 nM–1.0 mM). The proposed framework extracts electrochemical-based features and separately fuses, in the first step, impedimetric features, which are characteristic points and fixed frequency features, and the voltammetric features, which are current and potential features, for data reduction by LDA and classification by kNN. Then, in a second step, a decision fusion is carried out to combine the results for both measurement methods based on Dempster–Shafer (DS) evidence theory. The classification results reach an accuracy of 80.98% and 81.48% for voltammetric measurements and impedimetric measurements, respectively. The decision fusion based on DS evidence theory improves the total recognition accuracy to 91.60%, thus realizing significantly high accuracy in comparison to the state-of-the-art. In comparison, the feature fusion for both voltammetric and impedimetric features in one step reaches an accuracy of only 89.13%. The proposed hierarchical framework considers for the first time the fusion of impedimetric and voltammetric data and features from multiple electrochemical sensor arrays. The developed approach can be implemented for several further applications of pattern fusion, e.g., for electronic noses, measurement of environmental contaminants such as heavy metal ions, pesticides, explosives, and measurement of biomarkers, such as for the detection of cancers and diabetes.

Do Patient Outcomes Predict Loss to Long-Term Follow-Up After Spine Surgery?

To determine if spine surgery patients with greater improvement in patient-reported outcomes measures (PROMs) at early postoperative follow-up are more likely to be lost to follow-up at the 1-year and 2-year postoperative visits. All patients older than 18 years who underwent primary or revision decompression or fusion surgery for degenerative spinal conditions at an academic institution were retrospectively identified. Univariate analysis compared patient demographics, surgical factors, and changes in short-term and long-term postoperative PROMs (Neck Disability Index, Oswestry Disability Index, visual analog scale [VAS] neck, VAS arm, VAS back, VAS leg, and Short-Form 12 Physical and Mental Component Scores) across groups with and without 1 year and 2 years follow-up. Logistic regression isolated predictors of loss to follow-up. A total of 1412 patients were included. Younger patient age, primary surgery, and single surgical approach independently predicted loss at 1 year follow-up. Female sex predicted loss at 2 years follow-up, whereas multilevel fusion surgery predicted attendance at 2 years clinical follow-up. In patients lost at 1 year follow-up compared with those who attended, preoperative to 3-month Mental Component Score and VAS neck pain improvement was significantly greater. When comparing patients based on 2 years follow-up status, VAS back pain improvement at 1 year postoperatively was significantly greater in patients lost to 2 years follow-up. All other changes in PROMs did not differ significantly by 1 or 2 years follow-up attendance. Overall patient outcomes were not found to affect loss to long-term follow-up after spine surgery. The general lack of association between postoperative follow-up status and clinical outcome may limit bias introduced in retrospective PROM studies.

Finger Vein and Inner Knuckle Print Recognition Based on Multilevel Feature Fusion Network

Multimodal biometric recognition involves two critical issues: feature representation and multimodal fusion. Traditional feature representation requires complex image preprocessing and different feature-extraction methods for different modalities. Moreover, the multimodal fusion methods used in previous work simply splice the features of different modalities, resulting in an unsatisfactory feature representation. To address these two problems, we propose a Dual-Branch-Net based recognition method with finger vein (FV) and inner knuckle print (IKP). The method combines convolutional neural network (CNN), transfer learning, and triplet loss function to complete feature representation, thereby simplifying and unifying the feature-extraction process of the two modalities. Dual-Branch-Net also achieves deep multilevel fusion of the two modalities’ features. We assess our method on a public FV and IKP homologous multimodal dataset named PolyU-DB. Experimental results show that the proposed method performs best and achieves an equal error rate (EER) of the recognition result of 0.422%.

BgNet: Classification of benign and malignant tumors with MRI multi-plane attention learning.

To propose a deep learning-based classification framework, which can carry out patient-level benign and malignant tumors classification according to the patient's multi-plane images and clinical information. A total of 430 cases of spinal tumor, including axial and sagittal plane images by MRI, of which 297 cases for training (14072 images), and 133 cases for testing (6161 images) were included. Based on the bipartite graph and attention learning, this study proposed a multi-plane attention learning framework, BgNet, for benign and malignant tumor diagnosis. In a bipartite graph structure, the tumor area in each plane is used as the vertex of the graph, and the matching between different planes is used as the edge of the graph. The tumor areas from different plane images are spliced at the input layer. And based on the convolutional neural network ResNet and visual attention learning model Swin-Transformer, this study proposed a feature fusion model named ResNetST for combining both global and local information to extract the correlation features of multiple planes. The proposed BgNet consists of five modules including a multi-plane fusion module based on the bipartite graph, input layer fusion module, feature layer fusion module, decision layer fusion module, and output module. These modules are respectively used for multi-level fusion of patient multi-plane image data to realize the comprehensive diagnosis of benign and malignant tumors at the patient level. The accuracy (ACC: 79.7%) of the proposed BgNet with multi-plane was higher than that with a single plane, and higher than or equal to the four doctors' ACC (D1: 70.7%, p=0.219; D2: 54.1%, p<0.005; D3: 79.7%, p=0.006; D4: 72.9%, p=0.178). Moreover, the diagnostic accuracy and speed of doctors can be further improved with the aid of BgNet, the ACC of D1, D2, D3, and D4 improved by 4.5%, 21.8%, 0.8%, and 3.8%, respectively. The proposed deep learning framework BgNet can classify benign and malignant tumors effectively, and can help doctors improve their diagnostic efficiency and accuracy. The code is available at https://github.com/research-med/BgNet.

UPanGAN: Unsupervised pansharpening based on the spectral and spatial loss constrained Generative Adversarial Network

It is observed that, in most of the CNN-based pansharpening methods, the multispectral (MS) images are taken as the ground truth, and the downsampled panchromatic (Pan) and MS images are taken as the training data. However, the trained models from the downsampled images are not suitable for the pansharpening of the MS images with rich spatial and spectral information at their original spatial resolution. To tackle this problem, a novel iterative network based on spectral and textural loss constrained Generative Adversarial Network (GAN) is proposed for pansharpening. First, instead of directly outputting the fused imagery, the GAN focuses on generating the mean difference image. The input of the GAN is a good initial difference image, which will make the network work better. Second, the coarse-to-fine fusion framework is designed to generate the fused imagery. It uses two optimized discriminators to distinguish the generated images, and performs multi-level fusion processing on PAN and MS images to generate the best pansharpening image in full resolution. Finally, the well-designed loss functions are embedded into both the generator and the discriminators to accurately preserve the fidelity of the fused imagery. We validated our method by the images from QuickBird, GaoFen-2 and WorldView-2 satellites. The experimental results demonstrated that the proposed method obtained a better fusion performance than the state-of-the-art methods in both visual comparison and quantitative evaluation.

Safety and Efficacy of Recombinant Human Bone Morphogenetic Protein-2 in Multilevel Posterolateral Lumbar Fusion in a Prospective, Randomized, Controlled Trial.

ObjectiveThis study is an investigator-initiated, prospective, randomized, controlled study to evaluate the efficacy and safety of the combined use of recombinant human BMP-2 (rhBMP-2) and a hydroxyapatite (HA) carrier in multilevel fusion in patients with adult spinal deformity (ASD). MethodsThirty patients underwent posterolateral fusion for lumbar spinal deformities at 3 to 5 segments between L1 and S1. The patients received rhBMP-2+HA or HA on the left or right side of the transverse processes. They were followed up regularly at 1, 3, 6, and 12 months postoperatively. Fusion was defined according to the bone bridging on computed tomography scans. The fusion rate per segment was subanalyzed. Function and quality of life as well as pain in the lower back and lower extremities were evaluated. ResultsThe union rate for the rhBMP-2+HA group was 100% at 6 and 12 months. The union rate for the HA group was 77.8% (21 of 27) at 6 months and 88.0% (22 of 25) at 12 months (p = 0.014 at 6 months; not significant at 12 months). All segments were fused at 6 and 12 months in the rhBMP-2+HA group (p < 0.001). In the HA group, 108 of 115 segments (93.5%) were fused at 6 months and 105 of 109 segments (96.3%) at 12 months. Other clinical parameters (visual analogue scale, 36-item Short Form Health Survey, and Scoliosis Research Society-22 scores) improved compared to baseline. ConclusionCombining rhBMP-2 and an HA carrier is a safe and effective method to achieve multilevel fusion in patients with ASD.

Artifact-Assisted multi-level and multi-scale feature fusion attention network for low-dose CT denoising.

Since low-dose computed tomography (LDCT) images typically have higher noise that may affect accuracy of disease diagnosis, the objective of this study is to develop and evaluate a new artifact-assisted feature fusion attention (AAFFA) network to extract and reduce image artifact and noise in LDCT images. In AAFFA network, a feature fusion attention block is constructed for local multi-scale artifact feature extraction and progressive fusion from coarse to fine. A multi-level fusion architecture based on skip connection and attention modules is also introduced for artifact feature extraction. Specifically, long-range skip connections are used to enhance and fuse artifact features with different depth levels. Then, the fused shallower features enter channel attention for better extraction of artifact features, and the fused deeper features are sent into pixel attention for focusing on the artifact pixel information. Besides, an artifact channel is designed to provide rich artifact features and guide the extraction of noise and artifact features. The AAPM LDCT Challenge dataset is used to train and test the network. The performance is evaluated by using both visual observation and quantitative metrics including peak signal-noise-ratio (PSNR), structural similarity index (SSIM) and visual information fidelity (VIF). Using AAFFA network improves the averaged PSNR/SSIM/VIF values of AAPM LDCT images from 43.4961, 0.9595, 0.3926 to 48.2513, 0.9859, 0.4589, respectively. The proposed AAFFA network is able to effectively reduce noise and artifacts while preserving object edges. Assessment of visual quality and quantitative index demonstrates the significant improvement compared with other image denoising methods.

Risk Factors for Adjacent Segment Disease Following Anterior Cervical Discectomy and Fusion with Plate Fixation: A Systematic Review and Meta-Analysis.

Adjacent segment disease (ASD) following anterior cervical discectomy and fusion with plating (ACDF-P) may yield a poor prognosis or reoperation. This review aimed to summarize risk factors for radiographic ASD (RASD) and clinical ASD (CASD) after ACDF-P. Six electronic databases were searched from inception to October 30, 2021. Four reviewers independently screened titles, abstracts, and full-text articles to identify relevant studies. Methodological quality of the included studies was evaluated. Meta-analyses for risk factors were conducted, if possible. Sixteen cohort and 3 case-control studies (3,563 participants) were included. These studies showed low (n = 2), moderate (n = 9), and high (n = 8) risk of bias. One risk factor for RASD was pooled for 2 meta-analyses based on the follow-up period. Four different risk factors for CASD at ≥4 years were pooled for meta-analyses. Limited evidence showed that multi-level fusion, greater asymmetry in total or functional cross-sectional area of the cervical paraspinal muscle, and preoperative degeneration in a greater number of segments were associated with a higher RASD incidence <4 years after ACDF-P. In contrast, no significant risk factors were identified for CASD <4 years after ACDF-P. At ≥4 years after ACDF-P, limited evidence supported that both cephalad and caudal plate-to-disc distances of <5 mm were associated with a higher RASD incidence, and very limited evidence supported that developmental canal stenosis, preoperative RASD, unfused C5-C6 or C6-C7 adjacent segments, use of autogenous bone graft, and spondylosis-related ACDF-P were associated with a higher CASD incidence. Although several risk factors for RASD and CASD development after ACDF-P were identified, the supporting evidence was very limited to limited. Future prospective studies should extend the existing knowledge by more robustly identifying risk factors for RASD and CASD after ACDF-P to inform clinical practice. Prognostic Level IV. See Instructions for Authors for a complete description of levels of evidence.

Kinematics of a cervical disc prosthesis implanted above or below one- and two-level fusions

BackgroundThe theoretical advantages of hybrid constructs over multi-level fusion have been illustrated in clinical and biomechanical studies. However, there is no biomechanical data on hybrid constructs using load control analyses. There is also no clear data on whether there is a biomechanical difference if the arthroplasty is below or above a 1- or 2-level fusion. This work investigated the effect on segmental motion of having a cervical total disc arthroplasty implanted above or below a 1- or 2-level fusion. MethodsSegmental motions of 16 C2-T1 cervical spine specimens were measured as the specimens were tested to 1.5Nm in axial rotation and in flexion-extension under compressive preload. Tests were conducted on intact specimens, and then after arthroplasty with a 1-level and 2-level fusion. 8 specimens were in test Group 1, where the hybrid configuration had a total disc arthroplasty above a 1- or 2-level fusion. The arthroplasty was below the 1- and 2-level fusion in Group 2. Load control and displacement control analyses were conducted to determine the effect of the hybrid configurations on segmental motion. ResultsIn load control, compensatory motion increases were found at all non-instrumented cervical spine segments in flexion-extension and axial rotation. Flexion-extension and axial rotation ranges of motion at the total disc arthroplasty level were less than 1° different than intact.In displacement control, there was no consistent pattern of compensatory motion. Range of motion at the arthroplasty level was within 3.5° of intact. ConclusionsThe total disc arthroplasty segmental level in a hybrid construct has similar amounts of motion as intact. This may shield the arthroplasty level and adjacent levels from supra-physiological motion and loading. These results suggest that a hybrid construct may be protective of adjacent segments, whether the total disc arthroplasty is above or below the fusion.

HFFNet: hierarchical feature fusion network for blind binocular image quality prediction.

Compared with monocular images, scene discrepancies between the left- and right-view images impose additional challenges on visual quality predictions in binocular images. Herein, we propose a hierarchical feature fusion network (HFFNet) for blind binocular image quality prediction that handles scene discrepancies and uses multilevel fusion features from the left- and right-view images to reflect distortions in binocular images. Specifically, a feature extraction network based on MobileNetV2 is used to determine the feature layers from distorted binocular images; then, low-level binocular fusion features (or middle-level and high-level binocular fusion features) are obtained by fusing the left and right low-level monocular features (or middle-level and high-level monocular features) using the feature gate module; further, three feature enhancement modules are used to enrich the information of the extracted features at different levels. Finally, the total feature maps obtained from the high-, middle-, and low-level fusion features are applied to a three-input feature fusion module for feature merging. Thus, the proposed HFFNet provides better results, to the best of our knowledge, than existing methods on two benchmark datasets.

3D Vehicle Detection Using Multi-Level Fusion From Point Clouds and Images

3D vehicle detectors based on point clouds generally have higher detection performance than detectors based on multi-sensors. However, with the lack of texture information, point-based methods get many missing detection of occluded and distant vehicles, and false detection with high-confidence of similarly shaped objects, which is a potential threat to traffic safety. Therefore, in the long run, fusion-based methods have more potential. This paper presents a multi-level fusion network for 3D vehicle detection from point clouds and images. The fusion network includes three stages: data-level fusion of point clouds and images, feature-level fusion of voxel and Bird’s Eye View (BEV) in the point cloud branch, and feature-level fusion of point clouds and images. Besides, a novel coarse-fine detection header is proposed, which simulates the two-stage detectors, generating coarse proposals on the encoder, and refining them on the decoder. Extensive experiments show that the proposed network has better detection performance on occluded and distant vehicles, and reduces the false detection of similarly shaped objects, proving its superiority over some state-of-the-art detectors on the challenging KITTI benchmark. Ablation studies have also demonstrated the effectiveness of each designed module.

P43. Single-level posterior lumbar fusions in patients with Ehlers Danlos syndrome not found to be associated with increased postoperative adverse events or five-year reoperations

BACKGROUND CONTEXT Ehlers Danlos Syndrome (EDS) is a rare connective tissue disorder that results from mutations in collagen genes. Potentially related to laxity and resultant degenerative changes, adult EDS patients may require posterior lumbar fusion (PLF). However, with low numbers, adequately powered outcome studies have been limited. PURPOSE To investigate the effect of Ehlers Danlos Syndrome on postoperative outcomes and reoperations following PLF. STUDY DESIGN/SETTING Retrospective cohort study using the 2010 - 2020 MSpine Pearldiver dataset. PATIENT SAMPLE Patients with a diagnosis of EDS (any type) were identified using International Classification of Diseases (ICD) codes. Patients over 18 years of age undergoing PLF with or without laminectomy were identified using Current Procedural Terminology (CPT) codes. Patients who had a primary diagnosis code of fractures, infection, neoplasm or were not in the database for at least 90 days following the surgery were excluded. Further, patients who had multilevel fusions or had concurrent anterior fusion were excluded. OUTCOME MEASURES Ninety-day any, severe, and minor adverse events as well as readmissions and five-year reoperations. METHODS Adult patients undergoing single-level PLF (excluding any with anterior procedures) with and without EDS for which at least 90-day follow up was available were identified. Any cases performed for trauma, tumor, or infection were excluded. Single-level PLF EDS patients were then matched 1:4 with PLF non-EDS patients based on age, sex and Elixhauser Comorbidity Index (ECI). Rates of 90-day any, severe, and minor adverse events as well as readmissions were tabulated and compared with chi-square tests. Multivariate logistical regression was then performed (controlling for age, and ECI). Reoperation surgeries over five years were assessed, Kaplan-Meier survival curves generated, and curves of those with and without EDS were compared with log rank test. RESULTS In total, there were 170,100 single-level PLF case identified, of which 242 (0.14%) had EDS. After matching, there were 957 without EDS and 239 with EDS. On multivariate regression, there were no significant differences in 90-day any, severe, or minor adverse events, or readmissions (p>0.05 for each). Over five years, there were also not significant differences in rates of reoperation (p> 0.05). CONCLUSIONS For EDS patients undergoing PLF, the current study identified similar 90-day adverse events and five-year reoperation rates compared to those without EDS. These findings may be useful for patient counseling and surgical planning for those with this rare condition. FDA DEVICE/DRUG STATUS This abstract does not discuss or include any applicable devices or drugs. Ehlers Danlos Syndrome (EDS) is a rare connective tissue disorder that results from mutations in collagen genes. Potentially related to laxity and resultant degenerative changes, adult EDS patients may require posterior lumbar fusion (PLF). However, with low numbers, adequately powered outcome studies have been limited. To investigate the effect of Ehlers Danlos Syndrome on postoperative outcomes and reoperations following PLF. Retrospective cohort study using the 2010 - 2020 MSpine Pearldiver dataset. Patients with a diagnosis of EDS (any type) were identified using International Classification of Diseases (ICD) codes. Patients over 18 years of age undergoing PLF with or without laminectomy were identified using Current Procedural Terminology (CPT) codes. Patients who had a primary diagnosis code of fractures, infection, neoplasm or were not in the database for at least 90 days following the surgery were excluded. Further, patients who had multilevel fusions or had concurrent anterior fusion were excluded. Ninety-day any, severe, and minor adverse events as well as readmissions and five-year reoperations. Adult patients undergoing single-level PLF (excluding any with anterior procedures) with and without EDS for which at least 90-day follow up was available were identified. Any cases performed for trauma, tumor, or infection were excluded. Single-level PLF EDS patients were then matched 1:4 with PLF non-EDS patients based on age, sex and Elixhauser Comorbidity Index (ECI). Rates of 90-day any, severe, and minor adverse events as well as readmissions were tabulated and compared with chi-square tests. Multivariate logistical regression was then performed (controlling for age, and ECI). Reoperation surgeries over five years were assessed, Kaplan-Meier survival curves generated, and curves of those with and without EDS were compared with log rank test. In total, there were 170,100 single-level PLF case identified, of which 242 (0.14%) had EDS. After matching, there were 957 without EDS and 239 with EDS. On multivariate regression, there were no significant differences in 90-day any, severe, or minor adverse events, or readmissions (p>0.05 for each). Over five years, there were also not significant differences in rates of reoperation (p> 0.05). For EDS patients undergoing PLF, the current study identified similar 90-day adverse events and five-year reoperation rates compared to those without EDS. These findings may be useful for patient counseling and surgical planning for those with this rare condition.

P101. Can AI identify patterns of complex adult spinal deformity with distinct perioperative outcomes?

<h3>BACKGROUND CONTEXT</h3> Adult spinal deformity (ASD) refers to multiple types of spinal deformity. This study used a prospective multicenter database of patients with "complex" surgical ASD to derive a data-driven classification of different deformity patterns, and assessed if such patterns have distinct clinical outcomes. <h3>PURPOSE</h3> Use an AI-based unsupervised approach to identify patterns of ASD. <h3>STUDY DESIGN/SETTING</h3> Retrospective analysis of a multi-center prospective database. <h3>PATIENT SAMPLE</h3> A total of 286 patients with complex ASD who underwent surgical correction. <h3>OUTCOME MEASURES</h3> Operative decision-making and 30-day adverse events. <h3>METHODS</h3> Complex surgical ASD included severe deformity, surgical complexity or advanced age with a multilevel fusion. An unsupervised cluster analysis that allowed for 10% outliers was used to identify different patterns of deformity. Perioperative outcomes of these clusters were then compared using ANOVA, Kustal-Wallis, and Chi-Squared analyses as indicated, with p-value < 0.05 considered significant. <h3>RESULTS</h3> The Hyper-Kyphosis (Hyper TK, n = 31) group had a mean thoracic kyphosis of 82.6±17.6°, lumbar hyperextension (PI-LL: -20.4±14.7°) and relatively straight coronal curvatures. Hyper-Kyphosis patients were the youngest (mean age 48±20 years) and had the lowest disability (mean ODI 32.9±17.1) and pain scores (median NRS back 6 IQR 3 to 8, median NRS leg 1 IQR 0 to 4). The Severe Coronal (Coronal, n = 91) group had a mean thoracic Cobb of 44.7±17.8° and a mean thoraco-lumbar Cobb of 57.9±16.5° with preserved global sagittal alignment. Coronal patients had moderate disability (mean ODI 33.5±18.8), functional impairment (PCD: 34.4±12.3) and pain scores (median NRS back 7 IQR 5 to 8, median NRS leg 4 IQR 0 to 7). The Severe Sagittal (Sev. Sag. n=79) group had a severe spino-pelvic mismatch (42.9±12.9°) and global sagittal deformity (24.3±7.9°). Sev Sag had higher BMIs (28.9±5.9), high levels of disability (mean ODI 49.3±15.6) and low appearance scores (2.3±0.7). Finally, the Moderate Sagittal (Mod. Sag. n=85) group had moderate deformities without distinctive radiographic characteristics, a mean age of 68.8±7.8°, the highest PROMIS pain interference sub-scores (65.2±5.8), high levels of disability (mean ODI 47.6±15.3) and back pain (NRS back 7 IQR 6 to 9), and poor overall quality of life (mean SRS total 2.8±0.6). Thirty-day adverse events were equivalent. Hyper TK and Coronal patients had the longest constructs, but fusion to the pelvis was most common in Mod Sag (89.4%) and Sev Sag (97.5%) deformities. Coronal patients had more osteotomies per case (median 11 IQR 6.5 to 14), longer OR times, and more 30-day implant-related complications (5.5%). Sev Sag and Hyper TK patients were more likely to require a 3-column osteotomy (43% and 32.3%, respectively). Sev Sag deformities were more likely to require interbody implants (29.1%) and operative wound debridement (7.6%). Hyper TK patients had shorter hospital stays. <h3>CONCLUSIONS</h3> AI methodologies identified 4 distinct patient clusters within a large population of surgically treated ASD patients. Each ASD cluster presented with 1) a unique spinal deformity pattern, 2) reported distinct pathognomonic health deficits, 3) received consistent surgical treatment across 11 centers, and 4) had characteristic perioperative complications and hospital stays. <h3>FDA DEVICE/DRUG STATUS</h3> This abstract does not discuss or include any applicable devices or drugs.

153. In opioid naïve patients undergoing thoracolumbar spinal fusions, smoking increases odds for postoperative opioid use disorder

<h3>BACKGROUND CONTEXT</h3> Opioids are commonly used for management of pain following thoracolumbar spinal fusions. Numerous studies have identified patient and surgical risk factors for opioid dependence including preoperative opioid use, affective disorders, and invasiveness of surgery. However, the association of smoking and opioid use disorder, specifically in preoperative opioid naïve patients, has not been well characterized in spine literature. <h3>PURPOSE</h3> The purpose of this study is to characterize the risk for opioid use disorder (per DSM-V) after thoracolumbar spinal fusions among active smokers who were opioid naïve prior to surgery. <h3>STUDY DESIGN/SETTING</h3> This was a retrospective analysis using the Mariner-53 database (administrative database containing 53 million participants). <h3>PATIENT SAMPLE</h3> We sampled patients (age > 18 years) who were opioid naïve preoperatively and underwent thoracolumbar fusions with and without a history of smoking, based on ICD-9 and ICD-10 diagnostic codes. <h3>OUTCOME MEASURES</h3> Opioid utilization and rates of opioid use disorder within 6 months of surgery were measured and compared. <h3>METHODS</h3> A total of 30,856 patients were identified, and exact 1:1 matching based on baseline patient demographics, procedure type (single vs multi-level fusion), affective disorders (depression, anxiety), social determinants of health and comorbidities were used to create two groups with identical covariates: smoking group (n=2,960) and nonsmoking group (n=2,960). <h3>RESULTS</h3> Baseline characteristics were similar between both groups. Postoperatively, opioid utilization rates at 6 months were similar between both groups (56.8% vs 57.4%, p=0.65). However, opioid naïve patients who were active smokers had higher odds of developing opioid use disorder postoperatively compared to nonsmokers (OR 4.18, 95% CI: 3.20 - 5.46). <h3>CONCLUSIONS</h3> Active smokers who are opioid naïve prior to thoracolumbar spinal fusion are 4 times more likely to develop opioid use disorder postoperatively compared to nonsmokers. Further multi-institutional prospective studies are needed to corroborate our findings. <h3>FDA DEVICE/DRUG STATUS</h3> This abstract does not discuss or include any applicable devices or drugs.