Publications

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.

While physicians routinely consider uncertainty during patient diagnosis, large language models (LLMs) often fail to recognize that real-world clinical data can be too limited for a definitive diagnosis. Zhou et al. address this problem by training a LLM, ConfiDx, to recognize medical cases with limited clinical data. This approach improves the utility of LLMs in the clinic and enables physicians to more effectively recognize and explain uncertainty in their patient care.

Predicting clinical therapeutic outcomes from animal studies using conserved electrophysiological phenotypes could facilitate developing treatments for neuropsychiatric disorders. Alpha oscillations in human resting-state electroencephalogram recordings are altered in many disorders, but whether these disruptions exist in mouse models is unknown. Here, we employed a uniform analytical method to show in males with fragile X syndrome (FXS) that alpha oscillations in humans and alpha-like oscillations in the visual cortex of Fmr1-/y mice are slowed, with a stronger phenotype in adults than juveniles and a juvenile-specific power phenotype in both species. We find that alpha-like oscillations are disrupted by deletion of Fmr1 in cortical excitatory neurons and glia, reflect differential activity of two classes of GABAergic interneurons, and are more sensitive to activation of GABAB receptors by Arbaclofen in wild-type than Fmr1-/y mice. Our framework reveals evolutionary conservation of alpha oscillation disruptions, enables a deeper understanding of FXS pathophysiology, and narrows the gap between treatment promise and practice.

BACKGROUND: The long-term use of beta blockers after myocardial infarction in patients with preserved ventricular function is debated. General practitioners (GPs) often decide whether to continue or discontinue long-term medications, yet little is known about how they apply evolving evidence to clinical prescribing decisions.

OBJECTIVE: To assess whether GPs are willing to deprescribe beta blockers post myocardial infarction with preserved left ventricular function and to identify factors associated with deprescribing decisions.

DESIGN: Cross-sectional online survey using case vignettes, conducted between July 2023 and October 2024 in primary care settings in 24 sites across 20 European countries.

PARTICIPANTS: Practicing GPs recruited through convenience sampling at each site.

MAIN MEASURES: The primary outcome was whether the GP chose to deprescribe beta blockers in the vignettes. Adjusted risk ratios for the association between GP characteristics and the decision to deprescribe were estimated using Poisson regression with generalized estimating equations and robust standard errors, accounting for clustering at the GP and country level.

KEY RESULTS: 604 GPs participated in the survey (median [IQR] age, 44.0 [35.0-54.8] years; 364 [60.3%] female), 89.2% deprescribed beta blockers in at least one vignette. The likelihood of deprescribing increased with time since myocardial infarction (adjusted risk ratio [RR] = 1.28; 95% CI 1.21–1.36 after 5 years; RR = 1.78; 95% CI 1.66–1.90 after 10 years vs. 3 months) and with side effects (RR = 1.76; 95% CI 1.66–1.88). More years of clinical experience were associated with a lower likelihood of deprescribing (RR = 0.86; 95% CI 0.77–0.95 for most vs. least experienced).

CONCLUSIONS: In this cross-national vignette study, most GPs were willing to deprescribe beta blockers after myocardial infarction in patients with preserved left ventricular function, particularly when time had passed and side effects were present. These findings suggest that GPs are open to applying evolving evidence on beta blocker discontinuation in clinical care.

SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12875-026-03208-6.

Drusen and subretinal drusenoid deposits, the pathognomonic lesions for age-related macular degeneration (AMD), are rich in cholesterol. Yet, AMD is not consistently linked to plasma lipids. Here wild-type and human apolipoprotein B100-expressing (APOB100) mice were put on a Western type of diet for 13 months and then assessed for plasma lipid profile, high-density lipoprotein (HDL) heterogeneity, status of intraretinal and choroidal vasculatures, retinal structure, function, levels of cholesterol and other sterols, lipid and cholesterol distribution and expression of cholesterol-related genes. The dietary effects were more pronounced in APOB100 mice, which had human-like hyperlipidemia and different subpopulations of HDL3, than in wild-type mice. In addition, the APOB100 retina showed increased cholesterol input from the systemic circulation, higher cholesterol content, more cholesterol crystals, elevated expression of HDL-related genes, lipid accumulation in the retinal pigment epithelium and Bruch's membrane, and impaired function compared with the wild-type retina. Remarkably, in both genotypes, cholesterol crystals were detected in the choroid, piercing toward Bruch's membrane and leading to macrophage infiltration. Our data indicate how plasma lipid profile could be linked to AMD and that cholesterol crystals in the choroid should be further investigated as contributors to AMD development and progression.

BACKGROUND: Although widely employed in clinical research, no studies have evaluated the psychometric properties of the English version of the Borderline Symptom List–23 (BSL-23), including the equivalence of its mean score to that of the German version. Using Item Response Theory, we: a) examined the psychometric properties of the English BSL-23 when administered to treatment-seeking adults diagnosed with Borderline Personality Disorder (BPD); and b) tested if language-related bias needs to be considered when comparing or combining mean scores from the German and English versions.

METHOD: We evaluated the psychometric properties of the English BSL-23 by fitting a graded response model (GRM) to data from 321 treatment-seeking adults diagnosed with BPD. We used log-likelihood ratio tests to test for differential item functioning (DIF) between English and German items, fitting a multi-group GRM to data from the English sample and a sample of 327 German treatment-seeking adults diagnosed with BPD. We also tested for the cumulative effect of DIF on the mean score to assess potential bias in the mean score between the two versions.

RESULTS: The English BSL-23 mean score demonstrates high internal consistency(Cronbach’s α ≥ 0.90) across a wide range of BPD severity among treatment-seeking adults with BPD. Sensitivity of the English mean score to individual differences was greatest when BPD severity was mild to moderate. We found DIF at the item level but these effects cancelled out when items were summed, resulting in a positive bias in the mean score among English participants of no more than 0.10 (2.55%), possibly reflecting language differences.

CONCLUSION: The English BSL-23 mean score is highly reliable across a broad range of underlying BPD severity among treatment-seeking adults meeting 5 or more DSM-IV/DSM-5 criteria for BPD. Sensitivity is greatest when BPD severity is mild to moderate. Mean scores from the current versions of the English and German BSL-23, administered to treatment-seeking adults diagnosed with BPD, can be combined or compared without needing to adjust for language differences.

SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1186/s12888-026-07863-y.

INTRODUCTION: The Transparent Reporting of a multivariable prediction model of Individual Prognosis Or Diagnosis (TRIPOD) statement was published to improve the reporting and critical appraisal of prediction model studies for diagnosis and prognosis. This paper describes the processes and methods that will be used to develop an extension to the TRIPOD statement (TRIPOD-Code) for the management of code associated with prediction model studies. TRIPOD-Code focuses specifically on the transparent reporting of analytical code used in prediction model studies, including code for data preprocessing, model development, and model evaluation.

METHODS AND ANALYSIS: TRIPOD-Code will be developed following published guidance from the EQUATOR Network and will comprise five stages. Stage 1 will involve a methodological review of how code availability is reported in published prediction model studies. In Stage 2, we will consult a diverse group of key stakeholders using a Delphi process to identify items to be considered for inclusion in TRIPOD-Code. Stage 3 will consist of virtual consensus meetings to consolidate and prioritise the key items. Stage 4 will involve developing the TRIPOD-Code checklist. In the final stage, Stage 5, we will disseminate TRIPOD-Code via journals, conferences, blogs, websites (including TRIPOD and the EQUATOR Network) and social media. TRIPOD-Code will provide researchers working on prediction model studies with a reporting checklist and accompanying guidance to promote code completeness and availability.

ETHICS AND DISSEMINATION: This study has been determined to be exempt from ongoing IRB oversight by the Massachusetts Institute of Technology Committee on the Use of Humans as Experimental Subjects (COUHES) under Exempt ID: E-6675. Findings from this study will be disseminated through peer-reviewed publications.