A bstract: Alzheimer Disease (AD) poses a significant and growing public health challenge worldwide. Early and accurate diagnosis is crucial for effective intervention and care. In recent years, there has been a surge of interest in leveraging Electroen-cephalography (EEG) to improve the detection of AD. This paper focuses on the application of Graph Signal Processing (GSP) techniques using the Graph Discrete Fourier Transform (GDFT) to analyze EEG recordings for the detection of AD, by employing several machine learning (ML) and deep learning (DL) models. We evaluate our models on publicly available EEG data containing 88 patients categorized into three groups: AD, Frontotemporal Dementia (FTD), and Healthy Controls (HC). Binary classification of dementia versus HC reached a top accuracy of 85% (SVM), while multiclass classification of AD, FTD, and HC attained a top accuracy of 44% (Naive Bayes). We provide novel GSP methodology for detecting AD, and form a framework for further experimentation to investigate GSP in the context of other neurodegenerative diseases across multiple data modalities, such as neuroimaging data in Major Depressive Disorder, Epilepsy, and Parkinson disease.

Clinicians require a simple way to estimate prognosis after acute ischemic stroke (AIS). Existing models are complex and difficult to apply at bedside. We hypothesized that, after accounting for premorbid health, stroke severity predicts survival, and that the National Institutes of Health Stroke Scale (NIHSS) could serve as the basis for clinically intuitive mortality estimation. We conducted a retrospective ecological cohort study using the TriNetX Research Network, a federated database of electronic health records from more than 100 healthcare organizations. Adults (≥18 years) with AIS and a documented admission NIHSS were included. Propensity score matching aligned each NIHSS level with NIHSS= 0 on demographics and comorbidities. The primary outcome was 90-day mortality. Linear regression models, trained on odd NIHSS values and tested on even values, evaluated performance. Functional outcomes were analyzed with competing-risk methods. Among 147,391 eligible patients (median NIHSS=4), mortality increased linearly with severity: 90-day mortality (%) ≈ 1.7 × NIHSS. Model performance was strong (training R²=0.993; testing R²=0.986). Overall predicted and observed mortality were similar (13.3% vs 12.6%, P = .22), and discrimination was high (AUC=0.96). Functional outcomes displayed an inverted "u" pattern, driven by the competing risk of mortality at higher NIHSS levels. In this large ecological analysis, stroke severity predicted mortality after accounting for pre-stroke health. A single equation based solely on NIHSS accurately approximated 90-day mortality after AIS. While subject to the limitations of population-level inference, this "1.7% equation" offers a transparent, scalable framework that may support bedside communication and guide future prospective validation.

Background/Objective: Diabetes mellitus (DM) is a highly prevalent condition that contributes to adverse outcomes in patients undergoing total hip arthroplasty (THA). This study applied machine learning clustering algorithms to identify comorbidity profiles among diabetic THA patients and evaluate their association with postoperative outcomes. Methods: The 2015–2021 National Inpatient Sample was queried using ICD-10 CM/PCS codes to identify DM patients undergoing THA. Forty-nine comorbidities, complications, and clinical covariates were incorporated into clustering analysis. The Davies–Bouldin and Calinski–Harabasz indices determined the optimal number of clusters. Multivariate logistic regression assessed risk of non-routine discharge (NRD), and Kruskal–Wallis H testing evaluated length-of-stay (LOS) differences. Results: A total of 73,606 patients were included. Six clusters were identified, ranging from 107 to 61,505 patients. Cluster 6, enriched for urinary tract infection and sepsis, had the highest risk of NRD (OR 7.83, p < 0.001) and the longest median LOS (9.0 days). Clusters 1–4 had shorter recoveries with median LOS of 2.0 days and narrow variability, while Cluster 5 showed intermediate outcomes. Kruskal–Wallis and post hoc testing confirmed significant differences across clusters (p < 0.001). Conclusions: Machine learning clustering of diabetic THA patients revealed six distinct groups with varied comorbidity profiles. Infection-driven clusters carried the highest risk for non-routine discharge and prolonged hospitalization. This approach provides a novel framework for risk stratification and may inform targeted perioperative management strategies.

Background/Objective: Osteoarthritis (OA) is the most common indication for total hip arthroplasty (THA), yet postoperative utilization and discharge outcomes vary substantially due to heterogeneous comorbidity burdens. This study applied unsupervised machine learning clustering to identify distinct comorbidity profiles among OA patients undergoing THA and to evaluate their association with non-routine discharge (NRD) and length of stay (LOS). Methods: The 2015–2021 National Inpatient Sample was queried using ICD-10 CM/PCS codes to identify patients with OA undergoing THA. Forty-nine comorbidities, complications, and in-hospital clinical covariates were incorporated into a k-modes clustering analysis. The Davies–Bouldin and Calinski–Harabasz indices were used to determine the optimal number of clusters. Multivariable logistic regression assessed adjusted odds of NRD across clusters, and Kruskal–Wallis H testing evaluated differences in LOS. Results: A total of 401,846 patients were included, and five distinct clusters were identified, ranging from 777 to 331,755 patients. Clusters with higher prevalence of renal dysfunction, cardiovascular disease, anemia, and heart failure demonstrated significantly increased risk of NRD (adjusted odds ratios up to 3.01, p < 0.001) and prolonged hospitalization, with median LOS up to 4 days. Lower-risk clusters exhibited shorter hospitalizations with median LOS of 2 days and higher rates of routine discharge. Kruskal–Wallis testing confirmed significant LOS differences across all clusters (p < 0.001). Conclusions: Machine learning clustering of OA patients undergoing THA identified clinically distinct subgroups with graded differences in postoperative hospital utilization. Patients with greater comorbidity burden experienced disproportionately higher risk of NRD and prolonged LOS. This data-driven framework highlights heterogeneity within the OA population and may inform future strategies for perioperative risk stratification and resource planning.