Explainable Machine Learning Models for Prediction of Surgical Site Infection After Posterior Lumbar Fusion Surgery Based on Shapley Additive Explanations

OBJECTIVE

This study aims to develop machine learning (ML) models combined with an explainable method for the prediction of surgical site infection (SSI) after posterior lumbar fusion surgery.

METHODS

In this retrospective, single-center study, a total of 1016 consecutive patients who underwent posterior lumbar fusion surgery were included. A comprehensive dataset was established, encompassing demographic variables, comorbidities, preoperative evaluation, details related to diagnosed lumbar disease, preoperative laboratory tests, surgical specifics, and postoperative factors. Utilizing this dataset, 6nullML models were developed to predict the occurrence of SSI. Performance evaluation of the models on the testing set involved several metrics, including the receiver operating characteristic curve, the area under the receiver operating characteristic curve, accuracy, recall, F1 score, and precision. The Shapley Additive Explanations (SHAP) method was employed to generate interpretable predictions, enabling a comprehensive assessment of SSI risk and providing individualized interpretations of the model results.

RESULTS

Among the 1016 retrospective cases included in the study, 36 (3.54%) experienced SSI. Out of the six models examined, the Extreme Gradient Boost model demonstrated the highest discriminatory performance on the testing set, achieving the following metrics: precision (0.9000), recall (0.8182), accuracy (0.9902), F1 score (0.8571), and area under the receiver operating characteristic curve (0.9447). By utilizing the SHAP method, several important predictors of SSI were identified, including the duration of indwelling jugular vein catheter, blood urea nitrogen levels, total protein levels, sustained fever, creatinine levels, triglycerides levels, monocyte count, diabetes mellitus, drainage time, white blood cell count, cerebral infarction, estimated blood loss, prealbumin levels, Prognostic Nutritional Index, low back pain, posterior fusion score, and osteoporosis.

CONCLUSIONS

ML-based prediction tools can accurately assess the risk of SSI after posterior lumbar fusion surgery. Additionally, ML combined with SHAP could provide a clear interpretation of individualized risk prediction and give physicians an intuitive comprehension of the effects of the model's essential features.

Development and validation of a predictive model and tool for functional recovery in patients after postero-lateral interbody fusion

OBJECTIVE

The postoperative recovery of patients with lumbar disc herniation (LDH) requires further study. This study aimed to establish and validate a predictive model for functional recovery in patients with LDH and explore associated risk factors.

METHOD

Patients with LDH undergoing PLIF admitted from January 1, 2018 to December 31, 2022 were included, and patient data were prospectively collected through follow-up. The training and validation cohorts were randomly assigned in a 7:3 ratio. To pool data variables LASSO regression was used. The pooled variables were subsequently included in binary logistic regression analyses, construct risk prediction models, and plot nomograms. Additionally, recovery prediction models and interactive web page calculators were developed using R Shiny.

RESULTS

Overall, 1,097 patients with LDH following PLIF were included in this study. Regarding patients' economic and functional scores, 927 (84.5%) received excellent scores. Key indicators significantly were screened. Multivariate analysis showed that age, season, occupation, HDL-C, smoking, weekly exercise time, and osteoporosis were independent risk factors for postoperative recovery. The C-index of the model was 0.776 (95% CI: 0.7312-0.8208) and 0.804 (95% CI: 0.7408-0.8673) for the training and validation cohorts, respectively. The H-L test showed good fitting of the model (all P > 0.05). The DCA curve showed the best clinical efficacy when the threshold probability was in the ranges of 0-0.71 and 0.79-0.84. The interactive web calculator is accessed at https://postoperativerecoveryofldh.shinyapps.io/DynNomapp/ .

CONCLUSION

The predictive tools derived from this study can provide realistic and personalized expectations of postoperative outcomes for patients undergoing lumbar spine surgery.

Multicenter external validation of prediction models for clinical outcomes after spinal fusion for lumbar degenerative disease

BACKGROUND

Clinical prediction models (CPM), such as the SCOAP-CERTAIN tool, can be utilized to enhance decision-making for lumbar spinal fusion surgery by providing quantitative estimates of outcomes, aiding surgeons in assessing potential benefits and risks for each individual patient. External validation is crucial in CPM to assess generalizability beyond the initial dataset. This ensures performance in diverse populations, reliability and real-world applicability of the results. Therefore, we externally validated the tool for predictability of improvement in oswestry disability index (ODI), back and leg pain (BP, LP).

METHODS

Prospective and retrospective data from multicenter registry was obtained. As outcome measure minimum clinically important change was chosen for ODI with ≥ 15-point and ≥ 2-point reduction for numeric rating scales (NRS) for BP and LP 12 months after lumbar fusion for degenerative disease. We externally validate this tool by calculating discrimination and calibration metrics such as intercept, slope, Brier Score, expected/observed ratio, Hosmer-Lemeshow (HL), AUC, sensitivity and specificity.

RESULTS

We included 1115 patients, average age 60.8 ± 12.5 years. For 12-month ODI, area-under-the-curve (AUC) was 0.70, the calibration intercept and slope were 1.01 and 0.84, respectively. For NRS BP, AUC was 0.72, with calibration intercept of 0.97 and slope of 0.87. For NRS LP, AUC was 0.70, with calibration intercept of 0.04 and slope of 0.72. Sensitivity ranged from 0.63 to 0.96, while specificity ranged from 0.15 to 0.68. Lack of fit was found for all three models based on HL testing.

CONCLUSIONS

Utilizing data from a multinational registry, we externally validate the SCOAP-CERTAIN prediction tool. The model demonstrated fair discrimination and calibration of predicted probabilities, necessitating caution in applying it in clinical practice. We suggest that future CPMs focus on predicting longer-term prognosis for this patient population, emphasizing the significance of robust calibration and thorough reporting.

Predicting decompression surgery by applying multimodal deep learning to patients' structured and unstructured health data

BACKGROUND

Low back pain (LBP) is a common condition made up of a variety of anatomic and clinical subtypes. Lumbar disc herniation (LDH) and lumbar spinal stenosis (LSS) are two subtypes highly associated with LBP. Patients with LDH/LSS are often started with non-surgical treatments and if those are not effective then go on to have decompression surgery. However, recommendation of surgery is complicated as the outcome may depend on the patient's health characteristics. We developed a deep learning (DL) model to predict decompression surgery for patients with LDH/LSS.

MATERIALS AND METHOD

We used datasets of 8387 and 8620 patients from a prospective study that collected data from four healthcare systems to predict early (within 2 months) and late surgery (within 12 months after a 2 month gap), respectively. We developed a DL model to use patients' demographics, diagnosis and procedure codes, drug names, and diagnostic imaging reports to predict surgery. For each prediction task, we evaluated the model's performance using classical and generalizability evaluation. For classical evaluation, we split the data into training (80%) and testing (20%). For generalizability evaluation, we split the data based on the healthcare system. We used the area under the curve (AUC) to assess performance for each evaluation. We compared results to a benchmark model (i.e. LASSO logistic regression).

RESULTS

For classical performance, the DL model outperformed the benchmark model for early surgery with an AUC of 0.725 compared to 0.597. For late surgery, the DL model outperformed the benchmark model with an AUC of 0.655 compared to 0.635. For generalizability performance, the DL model outperformed the benchmark model for early surgery. For late surgery, the benchmark model outperformed the DL model.

CONCLUSIONS

For early surgery, the DL model was preferred for classical and generalizability evaluation. However, for late surgery, the benchmark and DL model had comparable performance. Depending on the prediction task, the balance of performance may shift between DL and a conventional ML method. As a result, thorough assessment is needed to quantify the value of DL, a relatively computationally expensive, time-consuming and less interpretable method.

A 9-year retrospective cohort of patients with lumbar disc herniation: Comparison of patient characteristics and recurrence frequency by smoking status

To evaluate the association between smoking status and patient characteristics and to identify risk factors associated with recurrence in patients who underwent surgery for lumbar disc herniation (LDH). This retrospective study was carried out at Lokman Hekim University, Ankara, Turkey between January 1, 2021 and January 1, 2022. The medical data of patients who underwent microsurgical discectomy for LDH were retrospectively recorded. Patients with any reemergence of LDH within a 6-month period after surgery were defined as having recurrent LDH. A total of 1109 patients were included in the study and mean age was 50.7 ± 14.3 years. The frequency of hernia at L2-L3 and L3-L4 levels was higher in the nonsmoker group (P < .001). The frequency of cases with Pfirrmann Grade 4 degeneration was higher in the nonsmoker group than in smokers and ex-smokers (P < .001). Protrusion-type hernias were more common in nonsmokers (P = .014), whereas paracentral hernias were more common in smokers (P < .001). The overall frequency of recurrence was 20.4%, and was higher in smokers than in non-smokers and ex-smokers (P < .001). Multivariable logistic regression revealed that current smoking (OR: 2.778, 95% CI [confidence interval]: 1.939-3.980, P < .001), presence of Pfirrmann Grade 4&5 disc degeneration (OR: 4.217, 95% CI: 2.966-5.996, P < .001), and paracentral herniation (OR: 5.040, 95% CI: 2.266-11,207, P < .001) were associated with higher risk of recurrence, whereas presence of sequestrated disc was associated with lower risk of recurrence (OR: 2.262, 95% CI:0.272-0.717, P = .001). Taken together, our data show that smoking, increased degree of degeneration and paracentral hernia increase the risk of LDH recurrence, while sequestrated disc appears to decrease risk. Taking steps to combat smoking in individuals followed for LDH may reduce the risk of recurrence in LDH patients.

Success and failure after surgery of degenerative disease of the lumbar spine: an operational definition based on satisfaction, pain, and disability from a prospective cohort

Background

To describe success and failure (S&F) after lumbar spine surgery in terms equally understandable across the entire health ecosystem.

Methods

Back and leg pain and disability were prospectively recorded before and up to 12 months after the procedure. Satisfaction was recorded using a Likert scale. Initially, patients were classified as satisfied or unsatisfied. Optimal satisfaction/unsatisfaction cutoff values for disability and pain were estimated with ROC curves. Satisfied and unsatisfied groups underwent a second subdivision into four subcategories: success (satisfied AND pain and disability concordant with cutoff values), incomplete success (satisfied AND pain and disability nonconformant with cutoff values), incomplete failure (unsatisfied AND pain and disability nonconformant with cutoff values), and failure (unsatisfied AND pain and disability concordant with cutoff values).

Results

A total of 486 consecutive patients were recruited from 2019–2021. The mean values of preoperative PROMs were ODI 42.2 (+ 16.4), NPRS back 6.6 (+ 2.6) and NPRS leg 6.2 points (+ 2.9). Of the total, 80.7% were classified as satisfied, and 19.3% were classified as unsatisfactory. The optimal disability and pain cutoff values for satisfaction/unsatisfaction (NPRS = 6, AND ODI = 27) defined a subdivision: 59.6% were classified as success, 20.4% as incomplete success, 7.1% as incomplete failure and 12.4% as failure. The descriptions of each group were translated to the following: success—all patients were satisfied and presented no or only mild to tolerable pain and no or borderline disability; incomplete success – all patients were satisfied despite levels of pain and/or disability worse than ideal for success; incomplete failure – all patients were not satisfied despite levels of pain and/or disability better than expected for failure; failure – all patients were unsatisfied and presented moderate to severe pain and disability.

Conclusion

It is possible to report S&F after surgery for DDL with precise and meaningful operational definitions focused on the experience of the patient.

Overweight and smoking promote recurrent lumbar disk herniation after discectomy

PURPOSE

Recurrent lumbar disk herniation (rLDH) following lumbar microdiscectomy is common. While several risk factors for primary LDH have been described, risk factors for rLDH have only sparsely been investigated. We evaluate the effect of Body mass index (BMI) and smoking on the incidence and timing of rLDH.

METHODS

From a prospective registry, we identified all patients undergoing primary tubular microdiscectomy (tMD), with complete BMI and smoking data, and a minimum 12-month follow-up. We defined rLDH as reherniation at the same level and side requiring surgery. Overweight was defined as BMI > 25, and obesity as BMI > 30. Intergroup comparisons and age- and gender-adjusted multivariable regression were carried out. We conducted a survival analysis to assess the influence of BMI and smoking on time to reoperation.

RESULTS

Of 3012 patients, 166 (5.5%) underwent re-microdiscectomy for rLDH. Smokers were reoperated more frequently (6.4% vs. 4.0%, p = 0.007). Similarly, rLDH was more frequent in obese (7.5%) and overweight (5.9%) than in normal-weight patients (3.3%, p = 0.017). Overweight smokers had the highest rLDH rate (7.6%). This effect of smoking (Odds ratio: 1.63, 96% CI: 1.12-2.36, p = 0.010) and BMI (Odds ratio: 1.09, 95% CI: 1.02-1.17, p = 0.010) persisted after controlling for age and gender. Survival analysis demonstrated that rLDH did not occur earlier in overweight patients and/or smokers.

CONCLUSIONS

BMI and smoking may directly contribute to a higher risk of rLDH, but do not accelerate rLDH development. Smoking cessation and weight loss in overweight or obese patients ought to be recommended with discectomy to reduce the risk for rLDH.

FUSE-ML: development and external validation of a clinical prediction model for mid-term outcomes after lumbar spinal fusion for degenerative disease

BACKGROUND

Indications and outcomes in lumbar spinal fusion for degenerative disease are notoriously heterogenous. Selected subsets of patients show remarkable benefit. However, their objective identification is often difficult. Decision-making may be improved with reliable prediction of long-term outcomes for each individual patient, improving patient selection and avoiding ineffective procedures.

METHODS

Clinical prediction models for long-term functional impairment [Oswestry Disability Index (ODI) or Core Outcome Measures Index (COMI)], back pain, and leg pain after lumbar fusion for degenerative disease were developed. Achievement of the minimum clinically important difference at 12 months postoperatively was defined as a reduction from baseline of at least 15 points for ODI, 2.2 points for COMI, or 2 points for pain severity.

RESULTS

Models were developed and integrated into a web-app ( https://neurosurgery.shinyapps.io/fuseml/ ) based on a multinational cohort [N = 817; 42.7% male; mean (SD) age: 61.19 (12.36) years]. At external validation [N = 298; 35.6% male; mean (SD) age: 59.73 (12.64) years], areas under the curves for functional impairment [0.67, 95% confidence interval (CI): 0.59-0.74], back pain (0.72, 95%CI: 0.64-0.79), and leg pain (0.64, 95%CI: 0.54-0.73) demonstrated moderate ability to identify patients who are likely to benefit from surgery. Models demonstrated fair calibration of the predicted probabilities.

CONCLUSIONS

Outcomes after lumbar spinal fusion for degenerative disease remain difficult to predict. Although assistive clinical prediction models can help in quantifying potential benefits of surgery and the externally validated FUSE-ML tool may aid in individualized risk-benefit estimation, truly impacting clinical practice in the era of "personalized medicine" necessitates more robust tools in this patient population.

Web-Based Calculator Predicts Surgical-Site Infection After Thoracolumbar Spine Surgery

BACKGROUND

Surgical-site infection (SSI) after spine surgery leads to increased length of stay, reoperation, and worse patient quality of life. We sought to develop a web-based calculator that computes an individual's risk of a wound infection following thoracolumbar spine surgery.

METHODS

We performed a retrospective review of consecutive patients undergoing elective degenerative thoracolumbar spine surgery at a tertiary-care institution between January 2016 and December 2018. Patients who developed SSI requiring reoperation were identified. Regression analysis was performed and model performance was assessed using receiver operating curve analysis to derive an area under the curve. Bootstrapping was performed to check for overfitting, and a Hosmer-Lemeshow test was employed to evaluate goodness-of-fit and model calibration.

RESULTS

In total, 1259 patients were identified; 73% were index operations. The overall infection rate was 2.7%, and significant predictors of SSI included female sex (odds ratio [OR] 3.0), greater body mass index (OR 1.1), active smoking (OR 2.8), worse American Society of Anesthesiologists physical status (OR 2.1), and greater surgical invasiveness (OR 1.1). The prediction model had an optimism-corrected area under the curve of 0.81. A web-based calculator was created: https://jhuspine2.shinyapps.io/Wound_Infection_Calculator/.

CONCLUSIONS

In this pilot study, we developed a model and simple web-based calculator to predict a patient's individualized risk of SSI after thoracolumbar spine surgery. This tool has a predictive accuracy of 83%. Through further multi-institutional validation studies, this tool has the potential to alert both patients and providers of an individual's SSI risk to improve informed consent, mitigate risk factors, and ultimately drive down rates of SSIs.

Predictive Model and Online Calculator for Discharge Disposition in Brain Tumor Patients

BACKGROUND

In the era of value-based payment models, it is imperative for neurosurgeons to eliminate inefficiencies and provide high-quality care. Discharge disposition is a relevant consideration with clinical and economic ramifications in brain tumor patients. We developed a predictive model and online calculator for postoperative non-home discharge disposition in brain tumor patients that can be incorporated into preoperative workflows.

METHODS

We reviewed all brain tumor patients at our institution from 2017 to 2019. A predictive model of discharge disposition containing preoperatively available variables was developed using stepwise multivariable logistic regression. Model performance was assessed using receiver operating characteristic curves and calibration curves. Internal validation was performed using bootstrapping with 2000 samples.

RESULTS

Our cohort included 2335 patients who underwent 2586 surgeries with a 16% non-home discharge rate. Significant predictors of non-home discharge were age >60 years (odds ratio [OR], 2.02), African American (OR, 1.73) or Asian (OR, 2.05) race, unmarried status (OR, 1.48), Medicaid insurance (OR, 1.90), admission from another health care facility (OR, 2.30), higher 5-factor modified frailty index (OR, 1.61 for 5-factor modified frailty index ≥2), and lower Karnofsky Performance Status (increasing OR with each 10-point decrease in Karnofsky Performance Status). The model was well calibrated and had excellent discrimination (optimism-corrected C-statistic, 0.82). An open-access calculator was deployed (https://neurooncsurgery.shinyapps.io/discharge_calc/).

CONCLUSIONS

A strongly performing predictive model and online calculator for non-home discharge disposition in brain tumor patients was developed. With further validation, this tool may facilitate more efficient discharge planning, with consequent improvements in quality and value of care for brain tumor patients.

Development of machine learning-based preoperative predictive analytics for unruptured intracranial aneurysm surgery: a pilot study

BACKGROUND

The decision to treat unruptured intracranial aneurysms (UIAs) or not is complex and requires balancing of risk factors and scores. Machine learning (ML) algorithms have previously been effective at generating highly accurate and comprehensive individualized preoperative predictive analytics in transsphenoidal pituitary and open tumor surgery. In this pilot study, we evaluate whether ML-based prediction of clinical endpoints is feasible for microsurgical management of UIAs.

METHODS

Based on data from a prospective registry, we developed and internally validated ML models to predict neurological outcome at discharge, as well as presence of new neurological deficits and any complication at discharge. Favorable neurological outcome was defined as modified Rankin scale (mRS) 0 to 2. According to the Clavien-Dindo grading (CDG), every adverse event during the post-operative course (surgery and not surgery related) is recorded as a complication. Input variables included age; gender; aneurysm complexity, diameter, location, number, and prior treatment; prior subarachnoid hemorrhage (SAH); presence of anticoagulation, antiplatelet therapy, and hypertension; microsurgical technique and approach; and various unruptured aneurysm scoring systems (PHASES, ELAPSS, UIATS).

RESULTS

We included 156 patients (26.3% male; mean [SD] age, 51.7 [11.0] years) with UIAs: 37 (24%) of them were treated for multiple aneurysm and 39 (25%) were treated for a complex aneurysm. Poor neurological outcome (mRS ≥ 3) was seen in 12 patients (7.7%) at discharge. New neurological deficits were seen in 10 (6.4%), and any kind of complication occurred in 20 (12.8%) patients. In the internal validation cohort, area under the curve (AUC) and accuracy values of 0.63-0.77 and 0.78-0.91 were observed, respectively.

CONCLUSIONS

Application of ML enables prediction of early clinical endpoints after microsurgery for UIAs. Our pilot study lays the groundwork for development of an externally validated multicenter clinical prediction model.

Machine learning-based preoperative predictive analytics for lumbar spinal stenosis

OBJECTIVEPatient-reported outcome measures (PROMs) following decompression surgery for lumbar spinal stenosis (LSS) demonstrate considerable heterogeneity. Individualized prediction tools can provide valuable insights for shared decision-making. The authors aim to evaluate the feasibility of predicting short- and long-term PROMs, reoperations, and perioperative parameters by machine learning (ML) methods.METHODSData were derived from a prospective registry. All patients had undergone single- or multilevel mini-open facet-sparing decompression for LSS. The prediction models were trained using various ML-based algorithms to predict the endpoints of interest. Models were selected by area under the receiver operating characteristic curve (AUC). The endpoints were dichotomized by minimum clinically important difference (MCID) and included 6-week and 12-month numeric rating scales for back pain (NRS-BP) and leg pain (NRS-LP) severity and the Oswestry Disability Index (ODI), as well as prolonged surgery (> 45 minutes), extended length of hospital stay (> 28 hours), and reoperations.RESULTSA total of 635 patients were included. The average age was 62 ± 10 years, and 333 patients (52%) were male. At 6 weeks, MCID was seen in 63%, 76%, and 61% of patients for ODI, NRS-LP, and NRS-BP, respectively. At internal validation, the models predicted MCID in these variables with accuracies of 69%, 76%, and 85%, and with AUCs of 0.75, 0.79, and 0.92. At 12 months, 66%, 63%, and 51% of patients reported MCID; the observed accuracies were 62%, 74%, and 66%, with AUCs of 0.68, 0.72, and 0.79. Reoperations occurred in 60 patients (9.5%), of which 27 (4.3%) occurred at the index level. Overall and index-level reoperations were predicted with 69% and 63% accuracy, respectively, and with AUCs of 0.66 and 0.61. In 15%, a length of surgery greater than 45 minutes was observed and predicted with 78% accuracy and AUC of 0.54. Only 15% of patients were admitted to the hospital for longer than 28 hours. The developed ML-based model enabled prediction of extended hospital stay with an accuracy of 77% and AUC of 0.58.CONCLUSIONSPreoperative prediction of a range of clinically relevant endpoints in decompression surgery for LSS using ML is feasible, and may enable enhanced informed patient consent and personalized shared decision-making. Access to individualized preoperative predictive analytics for outcome and treatment risks may represent a further step in the evolution of surgical care for patients with LSS.

The F2RaD Score: A Novel Prediction Score and Calculator Tool to Identify Patients at Risk of Postoperative C5 Palsy

BACKGROUND

Postoperative C5 palsy is a debilitating complication following posterior cervical decompression.

OBJECTIVE

To create a simple clinical risk score predicting the occurrence of C5 palsy

METHODS

We retrospectively reviewed all patients who underwent posterior cervical decompressions between 2007 and 2017. Data was randomly split into training and validation datasets. Multivariable analysis was performed to construct the model from the training dataset. A scoring system was developed based on the model coefficients and a web-based calculator was deployed.

RESULTS

The cohort consisted of 415 patients, of which 65 (16%) developed C5 palsy. The optimal model consisted of: mean C4/5 foraminal diameter (odds ratio [OR] = 9.1 for lowest quartile compared to highest quartile), preoperative C5 radiculopathy (OR = 3.5), and dexterity loss (OR = 2.9). The receiver operating characteristic yielded an area under the curve of 0.757 and 0.706 in the training and validation datasets, respectively. Every characteristic was worth 1 point except the lowest quartile of mean C4/5 foraminal diameter, which was worth 2 points, and the factors were summarized by the acronym F2RaD. The median predicted probability of C5 palsy increased from 2% in patients with a score of 0 to 70% in patients with a score of 4. The calculator can be accessed on https://jhuspine2.shinyapps.io/FRADscore/.

CONCLUSION

This study yielded a simplified scoring system and clinical calculator that predicts the occurrence of C5 palsy. Individualized risk prediction for patients may facilitate better understanding of the risks and benefits for an operation, and better prepare them for this possible adverse outcome. Furthermore, modifying the surgical plan in high-risk patients may possibly improve outcomes.

Prediction calculator for nonroutine discharge and length of stay after spine surgery

BACKGROUND CONTEXT

Following spine surgery, delays in referral to rehabilitation facilities leads to increased length of hospital stay (LOS), increases costs, more risk of hospital acquired complications, and decreased patient satisfaction.

PURPOSE

We sought to create a prediction calculator to determine the expected LOS after spine surgery and identify patients most likely to need postoperative nonhome discharge. The goal would be to facilitate earlier referral to rehabilitation and thereby ultimately shorten LOS, reduce costs, and improve patient satisfaction.

STUDY DESIGN

Retrospective.

PATIENT SAMPLE

We retrospectively reviewed all adult patients who underwent spine surgery for all indications between January and June 2018.

OUTCOME MEASURES

Length of stay and discharge disposition.

METHODS

Demographic variables, insurance status, baseline comorbidities, narcotic use, operative characteristics, as well as postoperative length of stay and discharge disposition data were collected. Univariable and multivariable analyses were performed to identify independent predictors of LOS and discharge disposition.

RESULTS

Two hundred fifty-seven patients were included. Mean age was 59 years, 46% were females, and 52% had private insurance vs 7% with Medicaid and 41% with Medicare. The most commonly performed procedure was lumbar fusion (31.9%). Mean LOS after surgery was 4.8 days and 18% had prolonged LOS >7 days. Age, insurance type, marriage status, and surgical procedure were significantly associated with LOS and discharge disposition. The final model had an area under the curve of 89% with good discrimination. A web based calculator was developed: https://jhuspine1.shinyapps.io/RehabLOS/ CONCLUSIONS: This study established a novel pilot calculator to identify those patients most likely to be discharged to rehabilitation facilities and to predict LOS after spine surgery. Our calculator had a high predictive accuracy of 89% compared to others in the literature. With validation this tool may ultimately facilitate streamlining of the postoperative period to shorten LOS, optimize resource utilization, and improve patient care.

Multicenter external validation of the Zurich Pituitary Score

PURPOSE

Recently, the Zurich Pituitary Score (ZPS) has been proposed as a new quantitative preoperative classification scheme for predicting gross total resection (GTR), extent of resection (EOR), and residual tumor volume (RV) in endoscopic pituitary surgery. We evaluated the external validity of the ZPS.

METHODS

In three reference centers for pituitary surgery, the ZPS was applied and correlated to GTR, EOR, and RV. Furthermore, its inter-rater agreement was assessed.

RESULTS

A total of 485 patients (53% male; age, 53.8 ± 15.7) were included. ZPS grades I, II, III, and IV were observed in 110 (23%), 270 (56%), 64 (13%), and 41 (8%) patients, respectively. GTR was achieved in 358 (74%) cases, with mean EOR of 87.6% ± 20.3% and RV of 1.42 ± 2.80 cm³. With increasing ZPS grade, strongly significant decreasing trends for GTR (I, 92%; II, 77%; III, 67%; IV, 15%; p < 0.001) and EOR (I, 93.8%; II, 89.9%; III, 88.1%; IV, 75.4%; p < 0.001) were found. Similarly, RV increased steadily ([cm³] I, 0.16; II, 0.61; III, 2.01; IV, 3.84; p < 0.001). We observed intraclass correlation coefficients of 0.837 (95% CI, 0.804-0.865) for intercarotid distance and 0.964 (95% CI, 0.956-0.970) for adenoma diameter, and Cohen's kappa of 0.972 (95% CI, 0.952-0.992) for the ZPS grades.

CONCLUSIONS

Application of the ZPS in three external cohorts was successful. The ZPS generalized well in terms of GTR, EOR, and RV; demonstrated excellent inter-rater agreement; and can safely and effectively be applied as a quantitative classification of adenomas with relevance to surgical outcome.

SpineCloud: image analytics for predictive modeling of spine surgery outcomes

Purpose: Data-intensive modeling could provide insight on the broad variability in outcomes in spine surgery. Previous studies were limited to analysis of demographic and clinical characteristics. We report an analytic framework called "SpineCloud" that incorporates quantitative features extracted from perioperative images to predict spine surgery outcome. Approach: A retrospective study was conducted in which patient demographics, imaging, and outcome data were collected. Image features were automatically computed from perioperative CT. Postoperative 3- and 12-month functional and pain outcomes were analyzed in terms of improvement relative to the preoperative state. A boosted decision tree classifier was trained to predict outcome using demographic and image features as predictor variables. Predictions were computed based on SpineCloud and conventional demographic models, and features associated with poor outcome were identified from weighting terms evident in the boosted tree. Results: Neither approach was predictive of 3- or 12-month outcomes based on preoperative data alone in the current, preliminary study. However, SpineCloud predictions incorporating image features obtained during and immediately following surgery (i.e., intraoperative and immediate postoperative images) exhibited significant improvement in area under the receiver operating characteristic (AUC): AUC = 0.72 ( CI 95 = 0.59 to 0.83) at 3 months and AUC = 0.69 ( CI 95 = 0.55 to 0.82) at 12 months. Conclusions: Predictive modeling of lumbar spine surgery outcomes was improved by incorporation of image-based features compared to analysis based on conventional demographic data. The SpineCloud framework could improve understanding of factors underlying outcome variability and warrants further investigation and validation in a larger patient cohort.

External validation of a prediction model for pain and functional outcome after elective lumbar spinal fusion

OBJECTIVE

Patient-reported outcome measures following elective lumbar fusion surgery demonstrate major heterogeneity. Individualized prediction tools can provide valuable insights for shared decision-making. We externally validated the spine surgical care and outcomes assessment programme/comparative effectiveness translational network (SCOAP-CERTAIN) model for prediction of 12-month minimum clinically important difference in Oswestry Disability Index (ODI) and in numeric rating scales for back (NRS-BP) and leg pain (NRS-LP) after elective lumbar fusion.

METHODS

Data from a prospective registry were obtained. We calculated the area under the curve (AUC), calibration slope and intercept, and Hosmer-Lemeshow values to estimate discrimination and calibration of the models.

RESULTS

We included 100 patients, with average age of 50.4 ± 11.4 years. For 12-month ODI, AUC was 0.71 while the calibration intercept and slope were 1.08 and 0.95, respectively. For NRS-BP, AUC was 0.72, with a calibration intercept of 1.02, and slope of 0.74. For NRS-LP, AUC was 0.83, with a calibration intercept of 1.08, and slope of 0.95. Sensitivity ranged from 0.64 to 1.00, while specificity ranged from 0.38 to 0.65. A lack of fit was found for all three models based on Hosmer-Lemeshow testing.

CONCLUSIONS

The SCOAP-CERTAIN tool can accurately predict which patients will achieve favourable outcomes. However, the predicted probabilities-which are the most valuable in clinical practice-reported by the tool do not correspond well to the true probability of a favourable outcome. We suggest that any prediction tool should first be externally validated before it is applied in routine clinical practice. These slides can be retrieved under Electronic Supplementary Material.

Deep learning-based preoperative predictive analytics for patient-reported outcomes following lumbar discectomy: feasibility of center-specific modeling

BACKGROUND CONTEXT

There is considerable variability in patient-reported outcome measures following surgery for lumbar disc herniation. Individualized prediction tools that are derived from center- or even surgeon-specific data could provide valuable insights for shared decision-making.

PURPOSE

To evaluate the feasibility of deriving robust deep learning-based predictive analytics from single-center, single-surgeon data.

STUDY DESIGN

Derivation of predictive models from a prospective registry.

PATIENT SAMPLE

Patients who underwent single-level tubular microdiscectomy for lumbar disc herniation.

OUTCOME MEASURES

Numeric rating scales for leg and back pain severity and Oswestry Disability Index scores at 12 months postoperatively.

METHODS

Data were derived from a prospective registry. We trained deep neural network-based and logistic regression-based prediction models for patient-reported outcome measures. The primary endpoint was achievement of the minimum clinically important difference (MCID) in numeric rating scales and Oswestry Disability Index, defined as a 30% or greater improvement from baseline. Univariate predictors of MCID were also identified using conventional statistics.

RESULTS

A total of 422 patients were included (mean [SD] age: 48.5 [11.5] years; 207 [49%] female). After 1 year, 337 (80%), 219 (52%), and 337 (80%) patients reported a clinically relevant improvement in leg pain, back pain, and functional disability, respectively. The deep learning models predicted MCID with high area-under-the-curve of 0.87, 0.90, and 0.84, as well as accuracy of 85%, 87%, and 75%. The regression models provided inferior performance measures for each of the outcomes.

CONCLUSIONS

Our study demonstrates that generating personalized and robust deep learning-based analytics for outcome prediction is feasible even with limited amounts of center-specific data. With prospective validation, the ability to preoperatively and reliably inform patients about the likelihood of symptom improvement could prove useful in patient counselling and shared decision-making.