Stanford Neurosurgical
Artificial Intelligence and Machine Learning Laboratory

The safety and efficacy of surgery is our number one priority. As we enter the era of big data, the onus is upon us to utilize what we have learned to improve medical care for future generations. Using novel, cutting edge artificial intelligence and machine learning techniques, our goal is to mine through millions of patient records to predict outcomes following all types of surgery. Imagine a virtual algorithm, that is capable of providing a true estimate of surgical risk, outcome, and efficacy – based specifically on your characteristics. This is the very foundation of precision medicine, and will empower physicians to deliver outstanding care.


The focus of my laboratory is to utilize precision medicine techniques to improve the diagnosis and treatment of neurologic conditions. From traumatic brain injury to spinal scoliosis, the ability to capture detailed data regarding clinical symptoms and treatment outcomes has empowered us to do better for patients. Utilize data to do better for patients, that’s what we do.

Associate Professor of Neurosurgery and, by courtesy, of Orthopaedic Surgery


  • Experience with the utilization of new-generation shared-control robotic system for spinal instrumentation. Journal of neurosurgical sciences Haider, G., Shah, V., Lopez, I., Wagner, K. E., Stienen, M. N., Veeravagu, A. 2024


    Robotic assistance in spine surgery is emerging as an accurate, effective and enabling technology utilized in the treatment of patients with surgical spinal pathology. The safety and reproducibility of robotic assistance in the placement of pedicle screw instrumentation is still being investigated. The objective of this study was to present our experience of instrumented spinal fusion utilizing an intraoperative robotic guidance system.We retrospectively reviewed all cases of spinal instrumentation of the thoracic and lumbo-sacral spine using the Mazor X robotic system (Medtronic Inc, Minneapolis, MN, USA), performed at our institution by one surgeon between July 2017 and June 2020. Wilcoxon Rank test was used to compare time taken to place each screw during the first 20 cases and the cases thereafter.A total of 28 patients were included. A total of 159 screws were placed using the Mazor X robotic system. The overall mean time for screw placement was 7.8±2.3 minutes and there was a significant reduction in the mean time for screw placement after the 20th case or 120 screws (8.70 vs. 5.42 min, P=0.008). No postoperative neurologic deficit or new radiculopathy was noted to occur secondary to hardware placement. No revision surgery was required for replacement or removal of a mispositioned screw.From this single-center, single-surgeon series we conclude that robot-assisted spine surgery can be safely and efficiently integrated into the operating room workflow, which improves after a learning curve of approximately 20 operative interventions. We found robot-assisted spinal instrumentation to be reliable, safe, effective and highly precise.

    View details for DOI 10.23736/S0390-5616.24.06206-4

    View details for PubMedID 38619188

  • Deep Learning Prediction of Cervical Spine Surgery Revision Outcomes Using Standard Laboratory and Operative Variables. World neurosurgery Schonfeld, E., Shah, A., Johnstone, T. M., Rodrigues, A., Morris, G. K., Stienen, M. N., Veeravagu, A. 2024


    INTRODUCTION: Cervical spine procedures represent a major proportion of all spine surgery. Mitigating the revision rate following cervical procedures requires careful patient selection. While complication risk has successfully been predicted, revision risk has proven more challenging. This is likely due to the absence of granular variables in claims databases. The objective of this study was to develop a state-of-the-art of revision prediction of cervical spine surgery using laboratory and operative variables.METHODS: Using the Stanford Research Repository, patients undergoing a cervical spine procedure between 2016-2022 were identified (N=3151) and recent laboratory values were collected. Patients were classified into separate cohorts by revision outcome and timeframe. Machine and deep learning models were trained to predict each revision outcome from laboratory and operative variables.RESULTS: Red blood cell count, Hemoglobin, Hematocrit, Mean Corpuscular Hemoglobin Concentration, Red Blood Cell Distribution Width, Platelet Count, CO2, Anion Gap, and Calcium were all significantly associated with one or more revision cohorts. For the prediction of 3-month revision, the deep neural network achieved AUC of 0.833. The model demonstrated increased performance for anterior than posterior and arthrodesis than decompression procedures.CONCLUSIONS: Our deep learning approach successfully predicted 3-month revision outcomes from demographic variables, standard laboratory values, and operative variables, in a cervical spine surgery cohort. This work introduces standard laboratory values and operative codes as meaningful predictive variables for revision outcome prediction. The increased performance on certain procedures evidences the need for careful development and validation of "one-size-fits-all" risk scores for spine procedures.

    View details for DOI 10.1016/j.wneu.2024.02.112

    View details for PubMedID 38408699

  • Misplaced intraspinal venous stent causing cauda equina syndrome: illustrative case. Journal of neurosurgery. Case lessons Shah, V., Johnstone, T., Haider, G., Marianayagam, N. J., Stienen, M. N., Chandra, V., Veeravagu, A. 2024; 7 (7)


    Endovenous stents for deep venous thrombosis treatment can be unintentionally placed in the spinal canal, resulting in neurological deficit.The authors report the case of a patient presenting to our institution with intraspinal misplacement of an endovenous stent, resulting in cauda equina syndrome. The authors also performed a systematic literature review, evaluating the few previously reported cases. This review was performed according to the updated Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. In four of five cases describing stent misplacement into the spinal canal, the authors report that only anteroposterior monoplanar imaging modalities were utilized for venous localization and stent deployment. The anteroposterior plane cannot assess the relative depth of structures, nor can it distinguish between superimposed structures well. Therefore, the use of biplanar imaging should at least be considered before stent deployment, as intraspinal stent placement can lead to disastrous consequences.This report should serve as an impetus for the use of biplanar or three-dimensional imaging modalities for iliac venous stent placement. Additionally, this work should increase spine surgeons' awareness about management and operative techniques when faced with this complication.

    View details for DOI 10.3171/CASE23482

    View details for PubMedID 38346298

  • Evaluating Computer Vision, Large Language, and Genome-Wide Association Models in a Limited Sized Patient Cohort for Pre-Operative Risk Stratification in Adult Spinal Deformity Surgery. Journal of clinical medicine Schonfeld, E., Pant, A., Shah, A., Sadeghzadeh, S., Pangal, D., Rodrigues, A., Yoo, K., Marianayagam, N., Haider, G., Veeravagu, A. 2024; 13 (3)


    Background: Adult spinal deformities (ASD) are varied spinal abnormalities, often necessitating surgical intervention when associated with pain, worsening deformity, or worsening function. Predicting post-operative complications and revision surgery is critical for surgical planning and patient counseling. Due to the relatively small number of cases of ASD surgery, machine learning applications have been limited to traditional models (e.g., logistic regression or standard neural networks) and coarse clinical variables. We present the novel application of advanced models (CNN, LLM, GWAS) using complex data types (radiographs, clinical notes, genomics) for ASD outcome prediction. Methods: We developed a CNN trained on 209 ASD patients (1549 radiographs) from the Stanford Research Repository, a CNN pre-trained on VinDr-SpineXR (10,468 spine radiographs), and an LLM using free-text clinical notes from the same 209 patients, trained via Gatortron. Additionally, we conducted a GWAS using the UK Biobank, contrasting 540 surgical ASD patients with 7355 non-surgical ASD patients. Results: The LLM notably outperformed the CNN in predicting pulmonary complications (F1: 0.545 vs. 0.2881), neurological complications (F1: 0.250 vs. 0.224), and sepsis (F1: 0.382 vs. 0.132). The pre-trained CNN showed improved sepsis prediction (AUC: 0.638 vs. 0.534) but reduced performance for neurological complication prediction (AUC: 0.545 vs. 0.619). The LLM demonstrated high specificity (0.946) and positive predictive value (0.467) for neurological complications. The GWAS identified 21 significant (p < 10-5) SNPs associated with ASD surgery risk (OR: mean: 3.17, SD: 1.92, median: 2.78), with the highest odds ratio (8.06) for the LDB2 gene, which is implicated in ectoderm differentiation. Conclusions: This study exemplifies the innovative application of cutting-edge models to forecast outcomes in ASD, underscoring the utility of complex data in outcome prediction for neurosurgical conditions. It demonstrates the promise of genetic models when identifying surgical risks and supports the integration of complex machine learning tools for informed surgical decision-making in ASD.

    View details for DOI 10.3390/jcm13030656

    View details for PubMedID 38337352

  • Getting What You Pay For: Impact of Copayments on Physical Therapy and Opioid Initiation, Timing, and Continuation for Newly Diagnosed Low Back Pain. The spine journal : official journal of the North American Spine Society Jin, M. C., Jensen, M., Barros Guinle, M. I., Ren, A., Zhou, Z., Zygourakis, C. C., Desai, A. M., Veeravagu, A., Ratliff, J. K. 2024


    Physical therapy (PT) is an important component of low back pain (LBP) management. Despite established guidelines, heterogeneity in medical management remains common.We sought to understand how copayments impact timing and utilization of PT in newly diagnosed LBP.The IBM Watson Health MarketScan claims database was utilized in a longitudinal setting.Adult patients with LBP.The primary outcomes-of-interest were timing and overall utilization of PT services. Additional outcomes-of-interest included timing of opioid prescribing.Actual and inferred copayments based on non-PCP visit claims were used to evaluate the relationship between PT copayment and incidence of PT initiation. Multivariable regression models were used to evaluate factors influencing PT usage.Overall, 2,467,389 patients were included. PT initiation, among those with at ≥1 PT service during the year after LBP diagnosis (30.6%), occurred at a median of 8 days post-diagnosis (IQR 1-55). Among those with at least one PT encounter, incidence of subsequent PT visits was significantly lower for those with high initial PT copayments. High initial PT copayments, while inversely correlated with PT utilization, were directly correlated with subsequent opioid use (0.77 prescriptions/patient [

Our Team

The Stanford Neurosurgical Artificial Intelligence and Machine Learning Laboratory is led by Dr. Anand Veeravagu, an Associate Professor of Neurosurgery and Associate Professor of Orthopedic Surgery, by courtesy, and Director of Minimally Invasive NeuroSpine Surgery at Stanford. Our laboratory team includes neurosurgery residents, clinical instructors, and medical students. 

We're Hiring!

Stanford Neurosurgical Artificial Intelligence and Machine Learning Laboratory is inviting applications for a 1-year post-doctoral research position. It offers an excellent opportunity for academic advancement and exposure to clinical neurosurgery. Responsibilities include clinical research productivity, database management, analytics, writing and study coordination. Highly motivated individuals with a medical degree, background analytics and prior neurosurgery experience welcomed. 

To apply, please contact us by email: