Outcomes from 871,441 Consecutive Surgical Procedures without Overlap or with Maximally Permissible Non-Concurrent Overlap.

OBJECTIVE: To isolate the impact of subsumed surgery (a shorter procedure completed entirely during overlapping non-critical portions of a longer antecedent procedure) on patient outcomes. SUMMARY BACKGROUND DATA: The American College of Surgeons recently recommended the elimination of "concurrent surgery" with overlap during a procedure's critical portions. Guidelines for non-concurrent overlap have been established, but the safety of subsumed surgery remains to be examined. METHODS: All consecutive procedures from 2013 to 2021 within a multihospital academic medical center were included (n=871,441). Simple logistic regression was performed to compare postoperative events between patients undergoing non-overlap surgery (n=533,032) and completely subsumed surgery (n=11,319). Thereafter, coarsened exact matching was used to match patients with non-overlap and subsumed surgery 1:1 on CPT code, 18 demographic features, baseline health characteristics, and procedural variables (n=7,146). Exact-matched cases were subsequently limited to pairs performed by the same surgeon (n=5,028). Primary outcomes included 30-day readmission, ED visits, and reoperations. RESULTS: Univariate analysis suggested that subsumed surgery had a higher 30-day risk of readmission (OR 1.55, P<0.0001), ED evaluation (OR 1.19, P<0.0001), and reoperation (OR 1.98, P<0.0001). When comparison was limited to the exact same procedure and patients were matched on demographics and health characteristics, there were no outcome differences between patients with subsumed surgery and non-overlapping surgery, even when limiting analyses to the same surgeon. CONCLUSIONS: Similar surgeries for similar patients result in similar outcomes whether there is completely subsumed or no overlap. Individual surgeons performing a specific procedure have no outcome differences with subsumed and non-overlapping cases.

Disparities in seizure outcomes revealed by large language models.

OBJECTIVE: Large-language models (LLMs) can potentially revolutionize health care delivery and research, but risk propagating existing biases or introducing new ones. In epilepsy, social determinants of health are associated with disparities in care access, but their impact on seizure outcomes among those with access remains unclear. Here we (1) evaluated our validated, epilepsy-specific LLM for intrinsic bias, and (2) used LLM-extracted seizure outcomes to determine if different demographic groups have different seizure outcomes. MATERIALS AND METHODS: We tested our LLM for differences and equivalences in prediction accuracy and confidence across demographic groups defined by race, ethnicity, sex, income, and health insurance, using manually annotated notes. Next, we used LLM-classified seizure freedom at each office visit to test for demographic outcome disparities, using univariable and multivariable analyses. RESULTS: We analyzed 84 675 clinic visits from 25 612 unique patients seen at our epilepsy center. We found little evidence of bias in the prediction accuracy or confidence of outcome classifications across demographic groups. Multivariable analysis indicated worse seizure outcomes for female patients (OR 1.33, P ≤ .001), those with public insurance (OR 1.53, P ≤ .001), and those from lower-income zip codes (OR ≥1.22, P ≤ .007). Black patients had worse outcomes than White patients in univariable but not multivariable analysis (OR 1.03, P = .66). CONCLUSION: We found little evidence that our LLM was intrinsically biased against any demographic group. Seizure freedom extracted by LLM revealed disparities in seizure outcomes across several demographic groups. These findings quantify the critical need to reduce disparities in the care of people with epilepsy.

Low Household Income Increases Hospital Length of Stay and Decreases Home Discharge Rates in Lumbar Fusion.

STUDY DESIGN: Retrospective Matched Cohort Study. OBJECTIVES: Low median household income (MHI) has been correlated with worsened surgical outcomes, but few studies have rigorously controlled for demographic and medical factors at the patient level. This study isolates the relationship between MHI and surgical outcomes in a lumbar fusion cohort using coarsened exact matching. METHODS: Patients undergoing single-level, posterior lumbar fusion at a single institution were consecutively enrolled and retrospectively analyzed (n = 4263). Zip code was cross-referenced to census data to derive MHI. Univariate regression correlated MHI to outcomes. Patients with low MHI were matched to those with high MHI based on demographic and medical factors. Outcomes evaluated included complications, length of stay, discharge disposition, 30- and 90 day readmissions, emergency department (ED) visits, reoperations, and mortality. RESULTS: By univariate analysis, MHI was significantly associated with 30- and 90 day readmission, ED visits, reoperation, and non-home discharge, but not mortality. After exact matching (n = 270), low-income patients had higher odds of non-home discharge (OR = 2.5, P = .016) and higher length of stay (mean 100.2 vs 92.6, P = .02). There were no differences in surgical complications, ED visits, readmissions, or reoperations between matched groups. CONCLUSIONS: Low MHI was significantly associated with adverse short-term outcomes from lumbar fusion. A matched analysis controlling for confounding variables uncovered longer lengths of stay and higher rates of discharge to post-acute care (vs home) in lower MHI patients. Socioeconomic disparities affect health beyond access to care, worsen surgical outcomes, and impose costs on healthcare systems. Targeted interventions must be implemented to mitigate these disparities.

Racial disparities in short-term spinal fusion outcomes across 4263 consecutive patients.

OBJECTIVE: Race plays a salient role in access to surgical care. However, few investigations have assessed the impact of race within surgical populations after care has been delivered. The objective of this study was to employ an exact matching protocol to a homogenous population of spine surgery patients in order to isolate the relationships between race and short-term postoperative outcomes. METHODS: In total, 4263 consecutive patients who underwent single-level, posterior-only lumbar fusion at a single multihospital academic medical center were retrospectively enrolled. Of these patients, 3406 patients self-identified as White and 857 patients self-identified as non-White. Outcomes were initially compared across all patients via logistic regression. Subsequently, White patients and non-White patients were exactly matched on the basis of key demographic and health characteristics (1520 matched patients). Outcome disparities were evaluated between the exact-matched cohorts. Primary outcomes were readmissions, emergency department (ED) visits, reoperations, mortality, intraoperative complications, and discharge disposition. RESULTS: Before matching, non-White patients were less likely to be discharged home and more likely to be readmitted, evaluated in the ED, and undergo reoperation. After matching, non-White patients experienced higher rates of nonhome discharge, readmissions, and ED visits. Non-White patients did not have more surgical complications either before or after matching. CONCLUSIONS: Between otherwise similar cohorts of spinal fusion cases, non-White patients experienced unfavorable discharge disposition and higher risk of multiple adverse postoperative outcomes. However, these findings were not accounted for by differences in surgical complications, suggesting that structural factors underlie the observed disparities.

Association Between Race and Short-Term Outcomes Across 3988 Consecutive Single-Level Spinal Fusions.

BACKGROUND AND OBJECTIVES: Race has implications for access to medical care. However, the impact of race, after access to care has been attained, remains poorly understood. The objective of this study was to isolate the relationship between race and short-term outcomes across patients undergoing a single, common neurosurgical procedure. METHODS: In this retrospective cohort study, 3988 consecutive patients undergoing single-level, posterior-only open lumbar fusion at a single, multihospital, academic medical center were enrolled over a 6-year period. Among them, 3406 patients self-identified as White, and 582 patients self-identified as Black. Outcome disparities between all White patients vs all Black patients were estimated using logistic regression. Subsequently, coarsened exact matching controlled for outcome-mitigating factors; White and Black patients were exact-matched 1:1 on key demographic and health characteristics (matched n = 1018). Primary outcomes included 30-day and 90-day hospital readmissions, emergency department (ED) visits, reoperations, mortality, discharge disposition, and intraoperative complication. RESULTS: Before matching, Black patients experienced increased rate of nonhome discharge, readmissions, ED visits, and reoperations (all P < .001). After exact matching, Black patients were less likely to be discharged to home (odds ratio [OR] 2.68, P < .001) and had higher risk of 30-day and 90-day readmissions (OR 2.24, P < .001; OR 1.91, P < .001; respectively) and ED visits (OR 1.79, P = .017; OR 2.09, P < .001). Black patients did not experience greater risk of intraoperative complication (unintentional durotomy). CONCLUSION: Between otherwise homogenous spinal fusion cohorts, Black patients experienced unfavorable short-term outcomes. These disparities were not explained by differences in intraoperative complications. Further investigation must characterize and mitigate institutional and societal factors that contribute to outcome disparities.

Quantifying a novel three-dimensional marker of scoliosis.

BACKGROUND: Scoliosis causes abnormal spinal curvature and torsional rotation of the vertebrae and has implications for human suffering and societal cost. In differential geometry, Writhe describes three-dimensional curvature. Differential geometric quantities can inform better diagnostic metrics of scoliotic deformity. This evaluation could help physicians and researchers study scoliosis and determine treatments. METHODS: Eight adult lumbar spine CT scans were analyzed in custom MATLAB programs to estimate Writhe and Cobb angle. Five patients exhibited scoliotic curvature, and three controls were asymptomatic. Vertebral centroids in three-dimensional space were determined, and Writhe was approximated. A T-test determined whether the affected spines had greater Writhe than the controls. Cohen's D test was used to determine effect size. RESULTS: Writhe of scoliotic spines (5.4E-4 ± 2.7E-4) was significantly higher than non-scoliotic spines (8.2E-5 ± 1.1E-4; p = 0.008). CONCLUSION: Writhe, a measure of curvature derived from 3D imaging, is significantly greater in scoliotic than in non-scoliotic spines. Future directions must include more subjects and examine writhe as a marker of scoliosis severity, progression, and response to treatment.

Disparities in seizure outcomes revealed by large language models.

Objective: Large-language models (LLMs) in healthcare have the potential to propagate existing biases or introduce new ones. For people with epilepsy, social determinants of health are associated with disparities in access to care, but their impact on seizure outcomes among those with access to specialty care remains unclear. Here we (1) evaluated our validated, epilepsy-specific LLM for intrinsic bias, and (2) used LLM-extracted seizure outcomes to test the hypothesis that different demographic groups have different seizure outcomes. Methods: First, we tested our LLM for intrinsic bias in the form of differential performance in demographic groups by race, ethnicity, sex, income, and health insurance in manually annotated notes. Next, we used LLM-classified seizure freedom at each office visit to test for outcome disparities in the same demographic groups, using univariable and multivariable analyses. Results: We analyzed 84,675 clinic visits from 25,612 patients seen at our epilepsy center 2005-2022. We found no differences in the accuracy, or positive or negative class balance of outcome classifications across demographic groups. Multivariable analysis indicated worse seizure outcomes for female patients (OR 1.33, p = 3×10-8), those with public insurance (OR 1.53, p = 2×10-13), and those from lower-income zip codes (OR ≥ 1.22, p ≤ 6.6×10-3). Black patients had worse outcomes than White patients in univariable but not multivariable analysis (OR 1.03, p = 0.66). Significance: We found no evidence that our LLM was intrinsically biased against any demographic group. Seizure freedom extracted by LLM revealed disparities in seizure outcomes across several demographic groups. These findings highlight the critical need to reduce disparities in the care of people with epilepsy.

The Impact of Neighborhood Socioeconomic Disadvantage on Operative Outcomes after Single-Level Lumbar Fusion.

INTRODUCTION: The relationship between socioeconomic status and neurosurgical outcomes has been investigated with respect to insurance status or median household income, but few studies have considered more comprehensive measures of socioeconomic status. This study examines the relationship between Area Deprivation Index (ADI), a comprehensive measure of neighborhood socioeconomic disadvantage, and short-term postoperative outcomes after lumbar fusion surgery. METHODS: 1861 adult patients undergoing single-level, posterior-only lumbar fusion at a single, multihospital academic medical center were retrospectively enrolled. An ADI matching protocol was used to identify each patient's 9-digit zip code and the zip code-associated ADI data. Primary outcomes included 30- and 90-day readmission, emergency department visits, reoperation, and surgical complication. Coarsened exact matching was used to match patients on key demographic and baseline characteristics known to independently affect neurosurgical outcomes. Odds ratios (ORs) were computed to compare patients in the top 10% of ADI versus lowest 40% of ADI. RESULTS: After matching (n = 212), patients in the highest 10% of ADI (compared to the lowest 40% of ADI) had significantly increased odds of 30- and 90-day readmission (OR = 5.00, P < 0.001 and OR = 4.50, P < 0.001), ED visits (OR = 3.00, P = 0.027 and OR = 2.88, P = 0.007), and reoperation (OR = 4.50, P = 0.039 and OR = 5.50, P = 0.013). There was no significant association with surgical complication (OR = 0.50, P = 0.63). CONCLUSIONS: Among otherwise similar patients, neighborhood socioeconomic disadvantage (measured by ADI) was associated with worse short-term outcomes after single-level, posterior-only lumbar fusion. There was no significant association between ADI and surgical complications, suggesting that perioperative complications do not explain the socioeconomic disparities in outcomes.

Quantifying interictal intracranial EEG to predict focal epilepsy.

Introduction: Intracranial EEG (IEEG) is used for 2 main purposes, to determine: (1) if epileptic networks are amenable to focal treatment and (2) where to intervene. Currently these questions are answered qualitatively and sometimes differently across centers. There is a need for objective, standardized methods to guide surgical decision making and to enable large scale data analysis across centers and prospective clinical trials. Methods: We analyzed interictal data from 101 patients with drug resistant epilepsy who underwent presurgical evaluation with IEEG. We chose interictal data because of its potential to reduce the morbidity and cost associated with ictal recording. 65 patients had unifocal seizure onset on IEEG, and 36 were non-focal or multi-focal. We quantified the spatial dispersion of implanted electrodes and interictal IEEG abnormalities for each patient. We compared these measures against the "5 Sense Score (5SS)," a pre-implant estimate of the likelihood of focal seizure onset, and assessed their ability to predict the clinicians' choice of therapeutic intervention and the patient outcome. Results: The spatial dispersion of IEEG electrodes predicted network focality with precision similar to the 5SS (AUC = 0.67), indicating that electrode placement accurately reflected pre-implant information. A cross-validated model combining the 5SS and the spatial dispersion of interictal IEEG abnormalities significantly improved this prediction (AUC = 0.79; p<0.05). The combined model predicted ultimate treatment strategy (surgery vs. device) with an AUC of 0.81 and post-surgical outcome at 2 years with an AUC of 0.70. The 5SS, interictal IEEG, and electrode placement were not correlated and provided complementary information. Conclusions: Quantitative, interictal IEEG significantly improved upon pre-implant estimates of network focality and predicted treatment with precision approaching that of clinical experts. We present this study as an important step in building standardized, quantitative tools to guide epilepsy surgery.

Generalization of finetuned transformer language models to new clinical contexts.

Objective: We have previously developed a natural language processing pipeline using clinical notes written by epilepsy specialists to extract seizure freedom, seizure frequency text, and date of last seizure text for patients with epilepsy. It is important to understand how our methods generalize to new care contexts. Materials and methods: We evaluated our pipeline on unseen notes from nonepilepsy-specialist neurologists and non-neurologists without any additional algorithm training. We tested the pipeline out-of-institution using epilepsy specialist notes from an outside medical center with only minor preprocessing adaptations. We examined reasons for discrepancies in performance in new contexts by measuring physical and semantic similarities between documents. Results: Our ability to classify patient seizure freedom decreased by at least 0.12 agreement when moving from epilepsy specialists to nonspecialists or other institutions. On notes from our institution, textual overlap between the extracted outcomes and the gold standard annotations attained from manual chart review decreased by at least 0.11 F1 when an answer existed but did not change when no answer existed; here our models generalized on notes from the outside institution, losing at most 0.02 agreement. We analyzed textual differences and found that syntactic and semantic differences in both clinically relevant sentences and surrounding contexts significantly influenced model performance. Discussion and conclusion: Model generalization performance decreased on notes from nonspecialists; out-of-institution generalization on epilepsy specialist notes required small changes to preprocessing but was especially good for seizure frequency text and date of last seizure text, opening opportunities for multicenter collaborations using these outcomes.

Relationship Between Comorbidity Burden and Short-Term Outcomes Across 4680 Consecutive Spinal Fusions.

OBJECTIVE: Preoperative management requires the identification and optimization of modifiable medical comorbidities, though few studies isolate comorbid status from related patient-level variables. This study evaluates Charlson Comorbidity Index (CCI)-an easily derived measure of aggregate medical comorbidity-to predict outcomes from spinal fusion surgery. Coarsened exact matching is employed to control for key patient characteristics and isolate CCI. METHODS: We retrospectively assessed 4680 consecutive patients undergoing single-level, posterior-only lumbar fusion at a single academic center. Logistic regression evaluated the univariate relationship between CCI and patient outcomes. Coarsened exact matching generated exact demographic matches between patients with high comorbid status (CCI >6) or no medical comorbidities (matched n = 524). Patients were matched 1:1 on factors associated with surgical outcomes, and outcomes were compared between matched cohorts. Primary outcomes included surgical complications, discharge status, 30- and 90-day risk of readmission, emergency department (ED) visits, reoperation, and mortality. RESULTS: Univariate regression of increasing CCI was significantly associated with non-home discharge, as well as 30- and 90-day readmission, ED visits, and mortality (all P < 0.05). Subsequent isolation of comorbidity between otherwise exact-matched cohorts found comorbid status did not affect readmissions, reoperations, or mortality; high CCI score was significantly associated with non-home discharge (OR = 2.50, P < 0.001) and 30-day (OR = 2.44, P = 0.02) and 90-day (OR = 2.29, P = 0.008) ED evaluation. CONCLUSIONS: Comorbidity, measured by CCI, did not increase the risk of readmission, reoperation, or mortality. Single-level, posterior lumbar fusions may be safe in appropriately selected patients regardless of comorbid status. Future studies should determine whether CCI can guide discharge planning and postoperative optimization.

Varying Degree of Overlap Following the Critical Steps of Lumbar Fusion and Short-term Outcomes.

STUDY DESIGN: Retrospective cohort. OBJECTIVE: The present study analyzes the impact of end-overlap on short-term outcomes after single-level, posterior lumbar fusions. SUMMARY OF BACKGROUND DATA: Few studies have evaluated how "end-overlap" (i.e., surgical overlap after the critical elements of spinal procedures, such as during wound closure) influences surgical outcomes. METHODS: Retrospective analysis was performed on 3563 consecutive adult patients undergoing single-level, posterior-only lumbar fusion over a 6-year period at a multi-hospital university health system. Exclusion criteria included revision surgery, missing key health information, significantly elevated body mass index (>70), non-elective operations, non-general anesthesia, and unclean wounds. Outcomes included 30-day emergency department visit, readmission, reoperation, morbidity, and mortality. Univariate analysis was carried out on the sample population, then limited to patients with end-overlap. Subsequently, patients with the least end-overlap were exact-matched to patients with the most. Matching was performed based on key demographic variables-including sex and comorbid status-and attending surgeon, and then outcomes were compared between exact-matched cohorts. RESULTS: Among the entire sample population, no significant associations were found between the degree of end-overlap and short-term adverse events. Limited to cases with any end-overlap, increasing overlap was associated with increased 30-day emergency department visits ( P =0.049) but no other adverse outcomes. After controlling for confounding variables in the demographic-matched and demographic/surgeon-matched analyses, no differences in outcomes were observed between exact-matched cohorts. CONCLUSIONS: The degree of overlap after the critical steps of single-level lumbar fusion did not predict adverse short-term outcomes. This suggests that end-overlap is a safe practice within this surgical population.

A Randomized Controlled Trial to Assess the Impact of Enhanced Recovery After Surgery on Patients Undergoing Elective Spine Surgery.

OBJECTIVE: To conduct a prospective, randomized controlled trial (RCT) of an enhanced recovery after surgery (ERAS) protocol in an elective spine surgery population. BACKGROUND: Surgical outcomes such as length of stay (LOS), discharge disposition, and opioid utilization greatly contribute to patient satisfaction and societal healthcare costs. ERAS protocols are multimodal, patient-centered care pathways shown to reduce postoperative opioid use, reduced LOS, and improved ambulation; however, prospective ERAS data are limited in spine surgery. METHODS: This single-center, institutional review board-approved, prospective RCT-enrolled adult patients undergoing elective spine surgery between March 2019 and October 2020. Primary outcomes were perioperative and 1-month postoperative opioid use. Patients were randomized to ERAS (n=142) or standard-of-care (SOC; n=142) based on power analyses to detect a difference in postoperative opioid use. RESULTS: Opioid use during hospitalization and the first postoperative month was not significantly different between groups (ERAS 112.2 vs SOC 117.6 morphine milligram equivalent, P =0.76; ERAS 38.7% vs SOC 39.4%, P =1.00, respectively). However, patients randomized to ERAS were less likely to use opioids at 6 months postoperatively (ERAS 11.4% vs SOC 20.6%, P =0.046) and more likely to be discharged to home after surgery (ERAS 91.5% vs SOC 81.0%, P =0.015). CONCLUSION: Here, we present a novel ERAS prospective RCT in the elective spine surgery population. Although we do not detect a difference in the primary outcome of short-term opioid use, we observe significantly reduced opioid use at 6-month follow-up as well as an increased likelihood of home disposition after surgery in the ERAS group.

Long-term epilepsy outcome dynamics revealed by natural language processing of clinic notes.

OBJECTIVE: Electronic medical records allow for retrospective clinical research with large patient cohorts. However, epilepsy outcomes are often contained in free text notes that are difficult to mine. We recently developed and validated novel natural language processing (NLP) algorithms to automatically extract key epilepsy outcome measures from clinic notes. In this study, we assessed the feasibility of extracting these measures to study the natural history of epilepsy at our center. METHODS: We applied our previously validated NLP algorithms to extract seizure freedom, seizure frequency, and date of most recent seizure from outpatient visits at our epilepsy center from 2010 to 2022. We examined the dynamics of seizure outcomes over time using Markov model-based probability and Kaplan-Meier analyses. RESULTS: Performance of our algorithms on classifying seizure freedom was comparable to that of human reviewers (algorithm F1 = .88 vs. human annotator κ = .86). We extracted seizure outcome data from 55 630 clinic notes from 9510 unique patients written by 53 unique authors. Of these, 30% were classified as seizure-free since the last visit, 48% of non-seizure-free visits contained a quantifiable seizure frequency, and 47% of all visits contained the date of most recent seizure occurrence. Among patients with at least five visits, the probabilities of seizure freedom at the next visit ranged from 12% to 80% in patients having seizures or seizure-free at the prior three visits, respectively. Only 25% of patients who were seizure-free for 6 months remained seizure-free after 10 years. SIGNIFICANCE: Our findings demonstrate that epilepsy outcome measures can be extracted accurately from unstructured clinical note text using NLP. At our tertiary center, the disease course often followed a remitting and relapsing pattern. This method represents a powerful new tool for clinical research with many potential uses and extensions to other clinical questions.

Neurosurgeons Deliver Similar Quality Care Regardless of First Assistant Type: Resident Physician versus Nonphysician Surgical Assistant.

OBJECTIVE: There are limited data evaluating the outcomes of attending neurosurgeons with different types of first assistants. This study considers a common neurosurgical procedure (single-level, posterior-only lumbar fusion surgery) and examines whether attending surgeons deliver equal patient outcomes, regardless of the type of first assistant (resident physician vs. nonphysician surgical assistant [NPSA]), among otherwise exact-matched patients. METHODS: The authors retrospectively analyzed 3395 adult patients undergoing single-level, posterior-only lumbar fusion at a single academic medical center. Primary outcomes included readmissions, emergency department visits, reoperation, and mortality within 30 and 90 days after surgery. Secondary outcome measures included discharge disposition, length of stay, and length of surgery. Coarsened exact matching was used to match patients on key demographics and baseline characteristics known to independently affect neurosurgical outcomes. RESULTS: Among exact-matched patients (n = 1402), there was no significant difference in adverse postsurgical events (readmission, emergency department visits, reoperation, or mortality) within 30 days or 90 days of the index operation between patients who had resident physicians and those who had NPSAs as first assistants. Patients who had resident physicians as first assistants demonstrated a longer length of stay (mean: 100.0 vs. 87.4 hours, P < 0.001) and a shorter duration of surgery (mean: 187.4 vs. 213.8 minutes, P < 0.001). There was no significant difference between the two groups in the percentage of patients discharged home. CONCLUSIONS: For single-level posterior spinal fusion, in the setting described, there are no differences in short-term patient outcomes delivered by attending surgeons assisted by resident physicians versus NPSAs.

Spike patterns surrounding sleep and seizures localize the seizure-onset zone in focal epilepsy.

OBJECTIVE: Interictal spikes help localize seizure generators as part of surgical planning for drug-resistant epilepsy. However, there are often multiple spike populations whose frequencies change over time, influenced by brain state. Understanding state changes in spike rates will improve our ability to use spikes for surgical planning. Our goal was to determine the effect of sleep and seizures on interictal spikes, and to use sleep and seizure-related changes in spikes to localize the seizure-onset zone (SOZ). METHODS: We performed a retrospective analysis of intracranial electroencephalography (EEG) data from patients with focal epilepsy. We automatically detected interictal spikes and we classified different time periods as awake or asleep based on the ratio of alpha to delta power, with a secondary analysis using the recently published SleepSEEG algorithm. We analyzed spike rates surrounding sleep and seizures. We developed a model to localize the SOZ using state-dependent spike rates. RESULTS: We analyzed data from 101 patients (54 women, age range 16-69). The normalized alpha-delta power ratio accurately classified wake from sleep periods (area under the curve = .90). Spikes were more frequent in sleep than wakefulness and in the post-ictal compared to the pre-ictal state. Patients with temporal lobe epilepsy had a greater wake-to-sleep and pre- to post-ictal spike rate increase compared to patients with extra-temporal epilepsy. A machine-learning classifier incorporating state-dependent spike rates accurately identified the SOZ (area under the curve = .83). Spike rates tended to be higher and better localize the seizure-onset zone in non-rapid eye movement (NREM) sleep than in wake or REM sleep. SIGNIFICANCE: The change in spike rates surrounding sleep and seizures differs between temporal and extra-temporal lobe epilepsy. Spikes are more frequent and better localize the SOZ in sleep, particularly in NREM sleep. Quantitative analysis of spikes may provide useful ancillary data to localize the SOZ and improve surgical planning.

Extracting seizure frequency from epilepsy clinic notes: a machine reading approach to natural language processing.

OBJECTIVE: Seizure frequency and seizure freedom are among the most important outcome measures for patients with epilepsy. In this study, we aimed to automatically extract this clinical information from unstructured text in clinical notes. If successful, this could improve clinical decision-making in epilepsy patients and allow for rapid, large-scale retrospective research. MATERIALS AND METHODS: We developed a finetuning pipeline for pretrained neural models to classify patients as being seizure-free and to extract text containing their seizure frequency and date of last seizure from clinical notes. We annotated 1000 notes for use as training and testing data and determined how well 3 pretrained neural models, BERT, RoBERTa, and Bio_ClinicalBERT, could identify and extract the desired information after finetuning. RESULTS: The finetuned models (BERTFT, Bio_ClinicalBERTFT, and RoBERTaFT) achieved near-human performance when classifying patients as seizure free, with BERTFT and Bio_ClinicalBERTFT achieving accuracy scores over 80%. All 3 models also achieved human performance when extracting seizure frequency and date of last seizure, with overall F1 scores over 0.80. The best combination of models was Bio_ClinicalBERTFT for classification, and RoBERTaFT for text extraction. Most of the gains in performance due to finetuning required roughly 70 annotated notes. DISCUSSION AND CONCLUSION: Our novel machine reading approach to extracting important clinical outcomes performed at or near human performance on several tasks. This approach opens new possibilities to support clinical practice and conduct large-scale retrospective clinical research. Future studies can use our finetuning pipeline with minimal training annotations to answer new clinical questions.