Machine-learning model predicting postoperative delirium in older patients using intraoperative frontal electroencephalographic signatures

Machine learning predicts spinal cord stimulation surgery outcomes and reveals novel neural markers for chronic pain

Spinal cord stimulation (SCS) is a well-accepted therapy for refractory chronic pain. However, predicting responders remain a challenge due to a lack of objective pain biomarkers. The present study applies machine learning to predict which patients will respond to SCS based on intraoperative electroencephalogram (EEG) data and recognized outcome measures. The study included 20 chronic pain patients who were undergoing SCS surgery. During intraoperative monitoring, EEG signals were recorded under SCS OFF (baseline) and ON conditions, including tonic and high density (HD) stimulation. Once spectral EEG features were extracted during offline analysis, principal component analysis (PCA) and a recursive feature elimination approach were used for feature selection. A subset of EEG features, clinical characteristics of the patients and preoperative patient reported outcome measures (PROMs) were used to build a predictive model. Responders and nonresponders were grouped based on 50% reduction in 3-month postoperative Numeric Rating Scale (NRS) scores. The two groups had no statistically significant differences with respect to demographics (including age, diagnosis, and pain location) or PROMs, except for the postoperative NRS (worst pain: p = 0.028; average pain: p < 0.001) and Oswestry Disability Index scores (ODI, p = 0.030). Alpha-theta peak power ratio differed significantly between CP3-CP4 and T3-T4 (p = 0.019), with the lowest activity in CP3-CP4 during tonic stimulation. The decision tree model performed best, achieving 88.2% accuracy, an F1 score of 0.857, and an area under the curve (AUC) of the receiver operating characteristic (ROC)  of 0.879. Our findings suggest that combination of subjective self-reports, intraoperatively obtained EEGs, and well-designed machine learning algorithms might be potentially used to distinguish responders and nonresponders. Machine and deep learning hold enormous potential to predict patient responses to SCS therapy resulting in refined patient selection and improved patient outcomes.

A New Canonical Log-Euclidean Kernel for Symmetric Positive Definite Matrices for EEG Analysis (Oct 2024)

OBJECTIVE

Working with the Riemannian manifold of symmetric positive-definite (SPD) matrices has become popular in electroencephalography (EEG) analysis. Frequently selected for its speed property is the manifold geometry provided by the log-euclidean Riemannian metric. However, the kernels used in the log-euclidean framework are not canonically based on the underlying geometry. Therefore, we introduce a new canonical log-euclidean (CLE) kernel.

METHODS

We used the log-euclidean metric tensor on the SPD manifold to derive the CLE kernel. We compared it with existing kernels, namely the affine-invariant, log-euclidean, and Gaussian log-euclidean kernel. For comparison, we tested the kernels on two paradigms: classification and dimensionality reduction. Each paradigm was evaluated on five open-access brain-computer interface datasets with motor-imagery tasks across multiple sessions. Performance was measured as balanced classification accuracy using a leave-one-session-out cross-validation. Dimensionality reduction performance was measured using AUClogRNX.

RESULTS

The CLE kernel itself is simple and easily turned into code, which is provided in addition to all the analytical solutions to relevant equations in the log-euclidean framework. The CLE kernel significantly outperformed existing log-euclidean kernels in classification tasks and was several times faster than the affine-invariant kernel for most datasets.

CONCLUSION

We found that adhering to the geometrical structure significantly improves the accuracy over two commonly used log-euclidean kernels while keeping the speed advantages of the log-euclidean framework.

SIGNIFICANCE

The CLE provides a good choice as a kernel in time-critical applications and fills a gap in the kernel methods of the log-euclidean framework.

Applying a transformer architecture to intraoperative temporal dynamics improves the prediction of postoperative delirium

BACKGROUND

Patients who experienced postoperative delirium (POD) are at higher risk of poor outcomes like dementia or death. Previous machine learning models predicting POD mostly relied on time-aggregated features. We aimed to assess the potential of temporal patterns in clinical parameters during surgeries to predict POD.

METHODS

Long short-term memory (LSTM) and transformer models, directly consuming time series, were compared to multi-layer perceptrons (MLPs) trained on time-aggregated features. We also fitted hybrid models, fusing either LSTM or transformer models with MLPs. Univariate Spearman's rank correlations and linear mixed-effect models establish the importance of individual features that we compared to transformers' attention weights.

RESULTS

Best performance is achieved by a transformer architecture ingesting 30 min of intraoperative parameter sequences. Systolic invasive blood pressure and given opioids mark the most important input variables, in line with univariate feature importances.

CONCLUSIONS

Intraoperative temporal dynamics of clinical parameters, exploited by a transformer architecture named TRAPOD, are critical for the accurate prediction of POD.

Artificial Intelligence in Surgery: A Systematic Review of Use and Validation

Background: Artificial Intelligence (AI) holds promise for transforming healthcare, with AI models gaining increasing clinical use in surgery. However, new AI models are developed without established standards for their validation and use. Before AI can be widely adopted, it is crucial to ensure these models are both accurate and safe for patients. Without proper validation, there is a risk of integrating AI models into practice without sufficient evidence of their safety and accuracy, potentially leading to suboptimal patient outcomes. In this work, we review the current use and validation methods of AI models in clinical surgical settings and propose a novel classification system. Methods: A systematic review was conducted in PubMed and Cochrane using the keywords "validation", "artificial intelligence", and "surgery", following PRISMA guidelines. Results: The search yielded a total of 7627 articles, of which 102 were included for data extraction, encompassing 2,837,211 patients. A validation classification system named Surgical Validation Score (SURVAS) was developed. The primary applications of models were risk assessment and decision-making in the preoperative setting. Validation methods were ranked as high evidence in only 45% of studies, and only 14% of the studies provided publicly available datasets. Conclusions: AI has significant applications in surgery, but validation quality remains suboptimal, and public data availability is limited. Current AI applications are mainly focused on preoperative risk assessment and are suggested to improve decision-making. Classification systems such as SURVAS can help clinicians confirm the degree of validity of AI models before their application in practice.

Intraoperative electroencephalogram patterns as predictors of postoperative delirium in older patients: a systematic review and meta-analysis

Background

Postoperative delirium (POD) significantly affects patient outcomes after surgery, leading to increased morbidity, extended hospital stays, and potential long-term cognitive decline. This study assessed the predictive value of intraoperative electroencephalography (EEG) patterns for POD in adults.

Methods

This systematic review and meta-analysis followed the PRISMA and Cochrane Handbook guidelines. A thorough literature search was conducted using PubMed, Medline, and CENTRAL databases focusing on intraoperative native EEG signal analysis in adult patients. The primary outcome was the relationship between the burst suppression EEG pattern and POD development.

Results

From the initial 435 articles identified, 19 studies with a total of 7,229 patients were included in the systematic review, with 10 included in the meta-analysis (3,705 patients). In patients exhibiting burst suppression, the POD incidence was 22.1% vs. 13.4% in those without this EEG pattern (p=0.015). Furthermore, an extended burst suppression duration associated with a higher likelihood of POD occurrence (p = 0.016). Interestingly, the burst suppression ratio showed no significant association with POD.

Conclusions

This study revealed a 41% increase in the relative risk of developing POD in cases where a burst suppression pattern was present. These results underscore the clinical relevance of intraoperative EEG monitoring in predicting POD in older patients, suggesting its potential role in preventive strategies.

Systematic Review Registration

This study was registered on International Platform for Registered Protocols for Systematic Reviews and Meta-Analyses: INPLASY202420001, https://doi.org/10.37766/inplasy2024.2.0001.

Electroencephalogram Biomarkers from Anesthesia Induction to Identify Vulnerable Patients at Risk for Postoperative Delirium

BACKGROUND

Postoperative delirium is a common complication in elderly patients undergoing anesthesia. Even though it is increasingly recognized as an important health issue, the early detection of patients at risk for postoperative delirium remains a challenge. This study aims to identify predictors of postoperative delirium by analyzing frontal electroencephalogram at propofol-induced loss of consciousness.

METHODS

This prospective, observational single-center study included patients older than 70 yr undergoing general anesthesia for a planned surgery. Frontal electroencephalogram was recorded on the day before surgery (baseline) and during anesthesia induction (1, 2, and 15 min after loss of consciousness). Postoperative patients were screened for postoperative delirium twice daily for 5 days. Spectral analysis was performed using the multitaper method. The electroencephalogram spectrum was decomposed in periodic and aperiodic (correlates to asynchronous spectrum wide activity) components. The aperiodic component is characterized by its offset (y intercept) and exponent (the slope of the curve). Computed electroencephalogram parameters were compared between patients who developed postoperative delirium and those who did not. Significant electroencephalogram parameters were included in a binary logistic regression analysis to predict vulnerability for postoperative delirium.

RESULTS

Of 151 patients, 50 (33%) developed postoperative delirium. At 1 min after loss of consciousness, postoperative delirium patients demonstrated decreased alpha (postoperative delirium: 0.3 μV2 [0.21 to 0.71], no postoperative delirium: 0.55 μV2 [0.36 to 0.74]; P = 0.019] and beta band power [postoperative delirium: 0.27 μV2 [0.12 to 0.38], no postoperative delirium: 0.38 μV2 [0.25 to 0.48]; P = 0.003) and lower spectral edge frequency (postoperative delirium: 10.45 Hz [5.65 to 15.04], no postoperative delirium: 14.56 Hz [9.51 to 16.65]; P = 0.01). At 15 min after loss of consciousness, postoperative delirium patients displayed a decreased aperiodic offset (postoperative delirium: 0.42 μV2 (0.11 to 0.69), no postoperative delirium: 0.62 μV2 [0.37 to 0.79]; P = 0.004). The logistic regression model predicting postoperative delirium vulnerability demonstrated an area under the curve of 0.73 (0.69 to 0.75).

CONCLUSIONS

The findings suggest that electroencephalogram markers obtained during loss of consciousness at anesthesia induction may serve as electroencephalogram-based biomarkers to identify at an early time patients at risk of developing postoperative delirium.

EDITOR’S PERSPECTIVE

Trail making test B in postoperative delirium: a replication study

Background

The Trail Making Test B (TMT-B) is indicative of cognitive flexibility and several other cognitive domains. Previous studies suggest that it might be associated with the risk of developing postoperative delirium, but evidence is limited and conflicting. We therefore aimed to replicate the association of preoperative TMT-B results with postoperative delirium.

Methods

We included older adults (≥65 yr) scheduled for major surgery and without signs of dementia to participate in this binational two-centre longitudinal observational cohort study. Presurgical TMT-B scores were obtained. Delirium was assessed twice daily using validated instruments. Logistic regression was applied and the area under the receiver operating characteristic curve calculated to determine the predictive performance of TMT-B. We subsequently included covariates used in previous studies for consecutive sensitivity analyses. We further analysed the impact of outliers, missing or impaired data.

Results

Data from 841 patients were included and of those, 151 (18%) developed postoperative delirium. TMT-B scores were statistically significantly associated with the incidence of postoperative delirium {odds ratio per 10-s increment 1.06 (95% confidence interval [CI] 1.02-1.09), P=0.001}. The area under the receiver operating characteristic curve was 0.60 ([95% CI 0.55-0.64], P<0.001). The association persisted after removing 21 outliers (1.07 [95% CI 1.03-1.07], P<0.001). Impaired or missing TMT-B data (n=88) were also associated with postoperative delirium (odds ratio 2.74 [95% CI 1.71-4.35], P<0.001).

Conclusions

The TMT-B was associated with postoperative delirium, but its predictive performance as a stand-alone test was low. The TMT-B alone is not suitable to predict delirium in a clinical setting.

Clinical trial registration

NCT02265263. (https://clinicaltrials.gov/ct2/show/results/NCT02265263).