Umaer Hanif

Background: Sleep questionnaires are widely used to screen for obstructive sleep apnea (OSA), but their incremental value beyond collecting basic demographics is rarely assessed. Using a large sample of sleep clinic intake questionnaire data that includes the Epworth Sleepiness Scale, questions on snoring and breathing pause as well as medical history data, we quantified questionnaire-item contributions relative to demographics for OSA severity classification. Methods: In a retrospective cohort of 67,397 adults referred for in-laboratory polysomnography and with questionnaire data, XGBoost classifiers were trained on all features (169 variables), questionnaire items only (161), or demographics only (age, gender, BMI, race; 4). Outcomes were the apnea–hypopnea index (AHI) and respiratory disturbance index (RDI) at 5, 15 and 30 events/h, evaluated on a held-out 10% test set; TabPFN v2.5 served as a secondary comparator. SHAP (SHapley Additive exPlanations) analysis was performed globally and within age and BMI strata; Berlin and STOP-Bang were not evaluated. Results: At AHI ≥5, ≥15 and ≥30, the demographics-only model reached area under the receiver operating characteristic curve (AUROC) 0.749, 0.718 and 0.726, 0.016–0.038 below the all-features model; across all six endpoints it retained 86–94% of the above-chance AUROC. The questionnaire-only model performed worst (AUROC 0.674–0.699 for AHI). Age, BMI and gender were the leading SHAP predictors at every threshold, and the contributions of age and BMI varied across strata. Four-class severity classification was poor (XGBoost accuracy ≤0.45). Conclusions: In this referred cohort, four demographic variables captured most of the achievable discrimination, and the questionnaire added at most 0.038 AUROC. Questionnaire-based screening tools should be benchmarked against a demographic baseline, and prospective validation is needed.

Study Objectives: To evaluate periodic limb movements during sleep (PLMS) in a large sleep disorders cohort and explore associations with sleep-disordered breathing (SDB), demographics and key clinical comorbidities. Methods: A retrospective analysis including 481,857 adults who completed overnight polysomnography (PSG) and a sleep/medical history questionnaire. Data collection spanned 240 sleep centers across 30 U.S. states. Periodic limb movement index (PLMI) was calculated from technician-scored PSGs, and all PSG measures were scored in accordance with American Academy of Sleep Medicine guidelines. Associations with PLMI were assessed using multivariable linear regression, with self-reported and objective measures analyzed separately, first in predictor-specific and then in simultaneous models. Results: The sample included 54.1% males, with a mean age of 54.4 ± 14.4 years (range: 18–107), average PLMI of 10.3 ± 22.1 events/hour and average Apnea Hypopnea Index (AHI) of 14.7 ± 19.1 events/hour. PLMI was highest among White individuals; lower among Hispanic, Asian, and Black individuals. PLMI was strongly negatively associated with SDB measures, persisting even when stratified by CPAP usage. In both predictor-specific and simultaneous multivariable models, PLMI was associated with diagnosed depression, insomnia score, elevated blood sugar and anemia. Objectively, PLMI was associated with arousal index and lower sleep efficiency. Conclusions: Our findings highlight the potential clinical relevance of PLMS across neuropsychiatric, cardiometabolic, and sleep domains. Racial disparities may reflect genetic and sociocultural factors. Most strikingly, in the setting of increased SDB, PLMS may be undercounted, potentially warranting further evaluation of scoring criteria. These results support further genetic and mechanistic investigations.

Objective. Automated sleep scoring offers the potential for fast, precise, and cost-effective analysis. This study evaluated the accuracy, robustness, and clinical agreement of four algorithms against manual polysomnography (PSG) in athletes, and assessed how signal degradation and individual physiological traits affect automated scoring in ambulatory settings. Approach. Data from 75 healthy athletes (91.1% male, mean age 26.3 ± 5.4 years, range [18–41], 204 PSG recordings) were retrospectively collected from studies conducted in our laboratory. Manual scoring was compared to four automated sleep staging algorithms: Luna, U-Sleep, Yet Another Spindle Algorithm (YASA), and Greifswald Sleep Stage Classifier (GSSC). Performance was evaluated on accuracy (per stage, macro and weighted F 1), sensitivity (confusion matrices), robustness (signal quality: R ² ; individual differences: beta coefficients) and clinical relevance (prediction of total sleep time (TST), sleep onset latency (SOL), wake after sleep onset (WASO) and sleep efficiency (SE)). Main results. All algorithms demonstrated strong agreement with manual scoring, with macro F 1 scores of 0.76 (GSSC), 0.71 (U-Sleep), 0.70 (YASA), and 0.69 (Luna). GSSC was the most accurate and robust across sport, physiological, behavioral, and anthropometric variables. It showed minimal bias and narrow limits of agreement for TST, WASO and SE. GSSC was also the most resilient to signal degradation (marginal R ² = 0.05), compared to YASA and Luna (0.06) and U-Sleep (0.24). Significance. Automated sleep staging algorithms demonstrated accuracy comparable to typical inter-rater agreement, supporting broader use in athletes. These algorithms could make PSG more accessible for both monitoring and research, providing deeper insights into sleep and its influence on athletic performance.

Background: Sleep is a potentially modifiable factor relevant to glycaemic control in type 1 diabetes (T1D), but evidence from real-world, multi-night assessments is limited. We investigated associations between objectively measured sleep and next-day glycaemic outcomes in children and adolescents with T1D.Methods: In this prospective observational study, 62 children and adolescents with T1D contributed 236 nights of home-based, simplified polysomnography sleep recordings with next-day continuous glucose monitoring and insulin pump data. Associations between nightly sleep characteristics and next-day outcomes were analysed using linear mixed-effects models adjusted for clinical and demographic covariates.Findings: Total sleep time showed a non-linear association with next-day glycaemic outcomes (global p=0·002). Compared with 6 h of sleep, time in range increased by 2·1–4·2 percentage points at 7–9 h, with reductions in time above range and mean sensor glucose. Duration of deep sleep (N3) was not associated with time in range (−1·6 percentage points per 30 min; 95% CI −3·6 to 0·4; p=0·12). Mean age was 12·8 years (SD 2·9), and 33 (53%) participants were female. Other sleep characteristics were associated with insulin pump–derived behavioural outcomes, particularly in adolescents.Interpretation: Total sleep time was the sleep characteristic most consistently associated with next-day glycaemic outcomes in pediatric T1D, with clinically relevant differences between 6 and 9 h of sleep. These findings support adequate sleep duration as a potentially modifiable target to improve day-to-day diabetes management and glycaemic outcomes.

Obstructive sleep apnea is a prevalent disorder associated with impaired quality of life and increased cardiometabolic risk, ultimately leading to heightened overall mortality. The first-line therapy is continuous positive airway pressure (CPAP), yet its effectiveness is limited by patient adherence. Follow-up protocols for CPAP management lack individualization, resulting in poor adherence and unnecessary costs. Artificial intelligence (AI) recently emerged as a promising strategy to address these limitations through enabling data-driven, individualized CPAP management. This narrative review synthesizes the extant evidence derived from 60 peer-reviewed studies published between 2017 and 2025 that applied AI to CPAP adherence. We focus on three complementary domains: (i) unsupervised learning methods to identify patient phenotypes and adherence trajectories, (ii) supervised models to predict short- and long-term adherence, and (iii) AI-enabled digital monitoring and intervention tools designed to improve sustained adherent CPAP use. AI-based approaches have consistently shown predictability, identifying early robust adherence patterns; these support the possibility of targeted, proactive interventions. Collectively, these findings suggest the need for a paradigm shift from “one size fits all” to personalized, behaviorally informed CPAP care. Future work should prioritize clinical validation, model interpretability, integration into care pathways, and real-world effectiveness to enable translation into routine clinical practice.

Diagnosing sleep disordered breathing requires manual annotation of events from sleep studies, such as nocturnal polysomnography, a process that is time-intensive, costly, and prone to inter-rater variability. Automatic approaches exist but lack generalizability due to signal variability across centers. We develop an automatic apneic breathing event detector to localize and classify obstructive apneas, central apneas, hypopneas, and isolated respiratory events without arousals or desaturations. The model is trained on 5456 polysomnographies and tested on 1099 polysomnographies from six cohorts uses an end-to-end deep learning architecture. The model’s predictions show a strong correlation with expert annotations for apnea-hypopnea index (r² = 0.84) and achieve an F1 score of 0.78 across apnea event types, with specific F1 scores of 0.71, 0.51, and 0.65 for obstructive apnea, central apnea, and hypopnea events, respectively. In two independent, multi-scored datasets, The model performs comparably or better than individual expert raters. The model’s probabilistic output, termed “apnotyping,” provides insights into sleep disordered breathing etiology, with event probabilities correlating more strongly with key sleep apnea traits—such as loop gain and pharyngeal muscle compensation—than traditional apnea indexes. This probabilistic approach may enhance diagnostic accuracy and support personalized treatment strategies, leading to improved patient outcomes.

Study Objectives: To evaluate the utility of the odds ratio product (ORP) in differentiating comorbid insomnia and sleep apnea (COMISA) from obstructive sleep apnea (OSA) and chronic insomnia (CIN). Methods: We retrospectively analyzed 9750 patients in four groups: (1) 1152 controls, (2) 2395 with CIN, (3) 2297 with OSA, and (4) 3906 with COMISA. CIN was defined as difficulty initiating/maintaining sleep with daytime fatigue/sleepiness occurring “often”/“always.” OSA was defined as an apnea-hypopnea index ≥5 on polysomnography. ORP, computed every 3 s from polysomnography, was analyzed alongside sleep metrics, comorbidities, and sleep habits. Associations were assessed using univariate multinomial logistic regression, followed by stepwise regression to identify independent predictors of COMISA versus OSA or CIN. Machine learning models classified COMISA, OSA, and CIN as distinct clinical groups. Results: ORP-derived features showed stronger associations with COMISA than traditional sleep metrics (except N3 latency). Independent objective predictors of COMISA included male sex (OR = 1.31, 95% CI = [1.16, 1.47]), BMI (1.27, [1.25, 1.29]), N3 latency (1.21, [1.13, 1.29]), age (1.17, [1.16, 1.19]), peak ORP during spontaneous arousals (1.12, [1.01, 1.25]), and time in ORP decile 7 (1.10, [1.07, 1.13]). Subjective predictors included depression, hypertension, allergy, headache, sleep aid/alcohol use, sleepiness, and lower sleep duration. Machine learning achieved overall accuracy of 61.2 per cent (p<.05), with sensitivity of 71 per cent for COMISA, 65 per cent for OSA, and 43 per cent for CIN. Conclusions: ORP is a promising objective marker for COMISA, distinguishing it from OSA more effectively than sleep metrics but separating COMISA from CIN poorly. Statement of Significance This study highlights the potential of the odds ratio product (ORP) as a novel objective marker to complement self-report questionnaires for identifying comorbid insomnia and sleep apnea (COMISA). Our findings demonstrate that ORP, combined with demographics and self-reported comorbidities and sleep habits, can distinguish COMISA from obstructive sleep apnea more effectively than traditional polysomnography metrics. These findings may refine the current understanding of COMISA from a research perspective and, in the future, may have clinical implications for improving the diagnosis and management of the disorder, which is often underrecognized in clinical patients who are evaluated for sleep-disordered breathing. Future studies should explore how ORP can be integrated into clinical sleep assessments and diagnostic strategies.

Polysomnography (PSG) is essential for diagnosing sleep disorders, but its manual interpretation is labor‐intensive. Automated sleep staging algorithms are promising, yet their utility in complex sleep disorders such as insomnia remains uncertain. This study evaluates five of the most recognised sleep staging classifiers—U‐Sleep, STAGES, GSSC, Luna and YASA—on PSG data from 904 patients with chronic insomnia. Performance was assessed using F1 scores, confusion matrices and predicted sleep metrics. The effect of demographics, sleepiness and PSG metrics on each classifier's performance was assessed using linear regression. Across all sleep stages, GSSC performed best (macro F1 score = 0.66), followed by U‐Sleep (0.62), Luna (0.56), STAGES (0.54) and YASA (0.52). GSSC achieved the highest F1 scores in Wake (0.83), N1 (0.22), N2 (0.80), N3 (0.71) and REM (0.76), while U‐Sleep matched its performance in N1 and REM and Luna in N3. STAGES performed poorest in N3 (0.39) and YASA in REM (0.35). Common misclassifications included N1 vs. Wake/N2 and N3 vs. N2, with REM misclassified as Wake/N1/N2 by STAGES, Luna and YASA. GSSC and U‐Sleep exhibited minimal demographic bias, while STAGES and Luna had more. No performance difference was observed between chronic insomnia patients with and without abnormal PSG. Sleep metric accuracy was highest for U‐Sleep (TST, R ² = 0.88), STAGES (SOL, R ² = 0.82) and GSSC (WASO, R ² = 0.82). These findings underscore the solid yet variable performance of the classifiers and highlight GSSC and U‐Sleep as leading tools for sleep staging in patients with chronic insomnia.

Understanding insomnia is crucial for improving its diagnosis and treatment. However, many subjective complaints about insomnia do not align with objective measures of sleep quality, as is the case in subjective-objective sleep discrepancy (SOSD). We address this discrepancy by measuring sleep intrusions and instability in polysomnographic recordings from a large clinical database. Using machine learning, we develop personalized models to infer hypnodensities—a continuous and probabilistic measure of sleep dynamics—, and analyze them via information theory to measure intrusions and instability in a principled way. We find that insomnia with SOSD involves sleep intrusions during intra-sleep wakefulness, while insomnia without SOSD shows wake intrusions during sleep, indicating distinct etiologies. By mapping these metrics to standard sleep features, we provide a continuous and interpretable framework for measuring sleep quality. This approach integrates and values subjective insomnia complaints with physiological data for a more accurate view of sleep quality and its disorders.