Publications

Interaural time and level differences are crucial in sound localization, yet their contributions to sound source segregation and spatial selection remain underspecified. Here, participants completed a spatial auditory selective attention task while we measured hemodynamic activity in the prefrontal cortex and superior temporal gyrus using functional near-infrared spectroscopy. Participants listened to a target sound stream and a simultaneous spatially separated speech stream or white noise masker. Sound streams were spatialized with either 50 μs interaural time differences (ITDs), 500 μs ITDs, naturally occurring interaural level differences (ILDs) from a non-individualized head-related transfer function (HRTF), or broadband 10 dB ILDs. Behavioral results revealed a stronger effect of spatial cues when the masker was speech. Error patterns differed in the two difficult conditions, small ITDs and natural ILDs: Small ITDs produced lower hit rates, while naturally occurring ILDs produced higher false alarm rates. Small ITDs led to greater activity in prefrontal cortex and activity in superior temporal gyrus that was lateralized, greater in the hemisphere contralateral to attentional focus, consistent with previous reports. These results suggest that natural ILDs alone support source segregation even if they are insufficient to cause large shifts in perceived lateralization, explaining high false alarm rates (confusions between target and distractor words). In contrast, small ITDs alone may be insufficient to segregate competing sources, leading to low hit and false alarm rates. Together, these results reveal differences in how ITDs and ILDs contribute to auditory scene analysis and spatial attention.

IMPORTANCE: The long-term efficacy of amyloid-targeting therapies hinges on their ability to slow downstream neuropathologic change, but little is known about the influence of amyloid clearance on tau pathology and neurodegeneration.

OBJECTIVE: To determine the postmortem and in vivo association between amyloid levels and downstream neuropathology after treatment with aducanumab in a patient with patchy areas showing minimal residual amyloid levels.

DESIGN, SETTING, AND PARTICIPANTS: This clinicopathologic case report from a single academic memory center includes a male carrier of the p.R47H TREM2 variant, which is associated with a higher risk of Alzheimer disease, who was in his 50s, had mild cognitive impairment, and received aducanumab while participating in a randomized clinical trial. Fourteen untreated controls, who were matched by age or presence of the TREM2 variant, also are included.

EXPOSURES: The male carrier of the p.R47H TREM2 variant had received 30 doses of aducanumab (cumulative dose of 280 mg/kg) over 4.5 years.

MAIN OUTCOMES AND MEASURES: Neuropathologic evaluation at autopsy, positron emission tomography to measure standardized uptake value ratio as a measure of amyloid and tau levels, and magnetic resonance imaging to determine longitudinal change in cortical thickness.

RESULTS: Four years after receiving the final dose of aducanumab, the patient died. An autopsy showed variable levels of amyloid pathology, including brain regions with very low levels of amyloid juxtaposed with brain regions that had typically high levels of amyloid in the deep cortical layers and only low levels of amyloid in the superficial cortical layers. Compared with the brain regions of the untreated controls, the brain regions of the patient after treatment with aducanumab showed low levels of amyloid that were preferentially found in the gyral crests, were associated with less tau pathology at autopsy, and were associated with slower longitudinal atrophy on in vivo magnetic resonance imaging (β = -0.50 [95% CI, -0.62 to -0.37]; t = -7.96 and P < .001). In contrast, the patient's brain regions with high amyloid burden were preferentially found in the sulcal depths and had similar levels of tau pathology as seen at autopsy in the untreated controls.

CONCLUSIONS AND RELEVANCE: In this case report, areas of extensive amyloid clearance after amyloid-targeting therapy were associated with less downstream neuropathologic change. In addition, amyloid clearance appears to preferentially occur in the gyral crests. Future studies should evaluate the differential mechanisms involved in amyloid clearance from superficial and deep cortical layers and in gyri and sulci because extensive amyloid clearance may be necessary to achieve downstream neuropathologic benefit after removal of amyloid.

INTRODUCTION: The impact of different neuropathologies on deep brain structures remains to be understood. We examine subcortical and limbic volumetry in neurodegenerative diseases involving phosphorylated tau (p-tau), α-synuclein, and transactive response DNA binding protein 43 (TDP-43).

METHODS: We acquired neuropathological measures and brain segmentations from postmortem analysis of 132 donors with Alzheimer's disease (AD), Lewy body disease (LBD), frontotemporal lobar degeneration with TDP-43 (FTLD-TDP), and FTLD-tau.

RESULTS: LBD had the least subcortical, limbic, and cortical atrophy compared to AD, FTLD-TDP, and FTLD-tau. In donors with both AD and LBD pathologies, primary LBD was associated with less atrophy than primary AD. While AD had cortico-subcortical and cortico-limbic morphometric associations, LBD had more limited parieto-occipital cortico-limbic associations. FTLD-TDP had cortico-subcortical while FTLD-tau had cortico-subcortical and cortico-limbic associations. In AD and FTLD-tau, hippocampal volumes correlated with p-tau burden, neuron loss, and gliosis. In LBD, thalamic α-synuclein severity was associated with subcortical/limbic volumes.

DISCUSSION: Postmortem neuroimaging reveals disease- and region-specific structure-pathology relationships.

BACKGROUND: Multiple Mini Interviews (MMIs) are widely used in medical school admissions to assess non-academic attributes. While MMIs are designed to promote fairness, questions remain about how applicant background may influence performance. In 2015, a single-institution UK study found no significant impact of school background or socioeconomic status on MMI scores. A decade later, this study re-examines those associations in a similar UK context.

METHODS: In this cross-sectional observational study, we analysed data from 545 home applicants to Imperial College School of Medicine during the 2025 admissions cycle. Using multiple linear regression, we explored the relationship between total MMI score and applicant gender, school type, and widening participation (WP) status. Station-level regressions were also conducted.

RESULTS: The regression model was significant and explained 7.6% of score variance. Out of a maximum score of 70, female applicants scored on average 3.2 points higher than males (95% CI: 2.1 to 4.2, p < 0.001). Applicants from independent and state-selective schools scored 2.4 (95% CI: 1.0 to 3.9, p = 0.001) and 1.3 points higher (95% CI: 0.0 to 2.7, p = 0.046), respectively, than state non-selective peers. WP status was not a significant predictor. Station-level analyses revealed no consistent disadvantage for WP applicants.

CONCLUSION: Our findings suggest that, unlike earlier UK research, school background and gender now show modest associations with MMI performance. WP applicants, however, perform comparably to their peers, indicating progress towards equitable access. Future work should explore the potential impact of coaching and familiarity with interview formats to inform MMI design.

IMPORTANCE: Emergency department (ED) quality review often uses administrative electronic triggers (eTriggers), but yields on detecting missed opportunities for diagnosis (MODs) are low. A commercial large language model (LLM) may help screen for MODs, yet evaluation data in real-world cohorts remain limited.

OBJECTIVE: To evaluate LLMs for identifying MODs in ED eTrigger cohorts.

DESIGN, SETTING, AND PARTICIPANTS: This retrospective diagnostic study of 2 eTrigger cohorts, ED discharge with return hospital admission within 72 hours and ED admission to the floor with intensive care unit (ICU) escalation within 24 hours, was conducted from April 2015 through March 2025 across 9 EDs (2 academic and 7 community) in 1 US health system. Samples included 200 encounters from the 72-hour return cohort and 100 encounters from the floor-to-ICU cohort; each case was adjudicated by 2 emergency physicians using a review process based on the Safer Dx framework.

EXPOSURES: Cases were evaluated by Claude Sonnet 4, Claude Sonnet 4.6, Claude Opus 4.6, Gemini 3 Pro, GPT-5, and GPT-5 mini.

MAIN OUTCOMES AND MEASURES: Main outcomes were sensitivity, specificity, positive predictive value, negative predictive value, area under the receiver operating characteristic curve (AUC), and reviewer-reviewer and reviewer-model concordance.

RESULTS: Among 300 sampled encounters, 12 were excluded, leaving 288 analyzed encounters (median [IQR] age, 69 [54-79] years; 135 female [46.9%]) with 39 MODs (13.5%), including 21 of 191 (11.0%) in the 72-hour return cohort and 18 of 97 (18.6%) in the floor-to-ICU cohort. Interrater agreement was 81.9% (95% CI, 77.4%-86.1%), with Gwet AC1 of 0.77 (95% CI, 0.70-0.83). In the 72-hour return cohort, model sensitivity ranged from 42.9% (95% CI, 24.5%-63.5%) for GPT-5 mini to 85.7% (95% CI, 65.4%-95.0%) for Claude Sonnet 4, specificity from 55.9% (95% CI, 48.4%-63.1%) for Claude Sonnet 4 to 82.9% (95% CI, 76.6%-87.9%) for GPT-5 mini, and AUC from 0.65 (95% CI, 0.53-0.77) for GPT-5 mini to 0.73 (95% CI, 0.61-0.85) for Claude Sonnet 4. In the floor-to-ICU cohort, sensitivity ranged from 5.6% (95% CI, 1.0-25.8%) for GPT-5 mini to 55.6% (95% CI, 33.7%-75.4%) for Claude Sonnet 4, specificity from 64.6% (53.6%-74.2%) for Claude Sonnet 4 to 97.5% (95% CI, 91.2%-99.3%) for GPT-5 mini, and AUC from 0.57 (95% CI, 0.46-0.67) for GPT-5 mini to 0.82 (95% CI, 0.73-0.91) for GPT-5. Across cohorts, LLMs showed similar discrimination but different sensitivity-specificity tradeoffs; Claude Sonnet 4 generally favored higher sensitivity, whereas GPT-5 mini favored higher specificity.

CONCLUSIONS AND RELEVANCE: In this diagnostic study of 2 ED eTrigger cohorts, model performance varied by cohort, with LLMs showing similar discrimination but different binary thresholds. These findings suggest that evaluation within the review workflow is needed before implementation and that reviewer-like concordance captures a distinct dimension of model behavior from discrimination.

Implicit bias and socioeconomic factors may influence diagnostic testing in pediatrics, but their effects on echocardiogram ordering in outpatient pediatric cardiology are not well understood. We evaluated whether patient demographics, insurance type, neighborhood Childhood Opportunity Index (COI), and provider characteristics (years of experience, degree attained, gender) were associated with guideline-concordant echocardiogram ordering for initial outpatient evaluation of pediatric chest pain. We conducted a retrospective study of 299 pediatric patients undergoing initial outpatient evaluation for chest pain between 2018 and 2024 at the University of Maryland Children's Hospital outpatient cardiology clinics. Echocardiogram appropriateness was determined using previously published appropriate use criteria (AUC) and structured clinical management and assessment plan (SCAMP) criteria. Associations between patient and provider factors and echocardiogram appropriateness were assessed. Echocardiogram ordering was concordant with AUC/SCAMP recommendations in 232 encounters (77.6%). White patients were more likely than non-White patients to receive guideline-concordant ordering (83.8% vs. 72.8%, p = .025). Patients from "Moderate", "Low", or "Very Low" COI neighborhoods were more likely to receive a non-indicated echocardiogram (27.5%, 31.3%, and 20%) compared to patients with "High" or "Very High" COI (16.4% and 14.4%) (p=.019). Providers with less than 10 years of experience were more likely to order echocardiograms in concordance with AUC/SCAMP criteria (86.9% vs. 71.2%, p < .001). Most echocardiogram ordering was guideline-concordant; however, deviations from guidelines were associated with patient race, neighborhood COI, and provider experience. Non-White patients and those from lower COI neighborhoods received more non-indicated echocardiograms than White patients or those from higher COI neighborhoods. These findings identify opportunities to improve equitable, guideline-concordant care in outpatient pediatric cardiology.

Shortly after its first outbreak in Wuhan, the SARS CoV-2 spread worldwide. This study determined SARS CoV-2 seroprevalence and compared anti-RBD IgG titre distributions among seropositive samples across four repeated cross-sectional surveys conducted during successive epidemic waves from August 2020 to January 2022. Age-specific seroprevalence and titre patterns were also examined across children, adolescents, and adults. We conducted a seroepidemiological cross-sectional study over four epidemic waves periods. A stratified 1,096 samples, reflecting the age and sex distribution of the Accra population, was selected from 15,000 residual clinical samples. Adjusted seroprevalence increased from 58.3% (95% CI: 53.9-62.6%) in wave 1 to > 90% by wave 2 and remained ≥ 96% in waves 3 and 4. Adults aged ≥ 20 years reached near-universal seropositivity earlier than children, who accumulated exposure more gradually. Despite this early plateau, geometric mean anti-RBD IgG concentrations increased from 162 BAU/mL in wave 1 to 625 BAU/mL in wave 3, stabilising at 584 BAU/mL in wave 4. Analyses of antibody titre distributions demonstrated that progressively larger proportions of adults showed concentrations substantially exceeding wave 1 baseline levels in later waves, indicating age-related differences in cumulative antigenic exposure. Across successive epidemic waves, Accra experienced intense and repeated SARS CoV-2 transmission, resulting in near-universal exposure of the adult population by May 2021. Children and adolescents showed a slower and more gradual increase in both seroprevalence and antibody concentrations. Anti-RBD IgG titres increased across successive waves and stabilised after August-November 2021, consistent with sustained cumulative antigenic exposure at the population level.

STUDY OBJECTIVES: To compare end-of-life predictions as measured by the physician-answered surprise question (SQ), "Would you be surprised if this patient died in the next 6 months?"), the Geriatric End-of-Life Screening Tool (GEST) artificial intelligence (AI) model, and a new collaborative GEST+SQ model for predicting 6-month mortality in older emergency department (ED) patients.

METHODS: This was a single-site prospective cohort study (Nov 2022 to June 2023) at a tertiary academic ED of patients aged 65 years and older. Answers to the SQ were collected within the electronic health record at ED disposition and GEST scores were calculated from available records using laboratory, vital signs, demographic and historical data. Six-month mortality was adjudicated via electronic health record and state records. SQ and GEST were compared using sensitivity and specificity. A new logistic regression model was developed combining SQ and GEST (GEST+SQ) and compared with GEST alone, using area under receiver-operating characteristic curves (ROC-AUC) for discrimination and expected calibration error for calibration. We modeled a sequential screening pathway where low- and high-risk patients received only GEST screening, whereas intermediate-risk patients received both GEST and SQ, reporting the proportion of patients for whom adding the SQ to GEST would change a theoretical referral to intervention.

RESULTS: From 9,256 eligible patients, 3,479 had SQ responses (37.6%), with 13.3% 6-month mortality. When matching GEST sensitivity to SQ (83.8%), GEST had greater specificity than the SQ (61.5% [56.7 to 67.1] vs. 50.8% [49.1 to 52.6]). At matching specificity (50.8%), GEST sensitivity (90.0% [87.0 to 92.7]) exceeded the SQ (83.8% [80.3 to 87.0]). GEST had an receiver-operating characteristic - area under the curve (ROC-AUC) of 0.79 (0.77 to 0.81), whereas the GEST+SQ model had ROC-AUC of 0.80 (0.78 to 0.82). The GEST+SQ model had significantly improved expected calibration error of 0.01 (0.01 to 0.02) for GEST+SQ vs. 0.042 (0.03 to 0.05) for GEST alone. In a sequential screening pathway, as few as 5% of patients required SQ screening following GEST risk scoring.

CONCLUSION: GEST modestly outperformed the SQ for predicting 6-month mortality. A GEST+SQ collaborative model did not improve discrimination (ROC-AUC) over GEST alone, but improved calibration. Sequential screening using GEST and then the SQ for intermediate-risk patients could decrease physician screening burden by 95% relative to manual, SQ-only screening. Collaborative approaches integrating automated tools with targeted physician input may enhance ED mortality risk assessment while reducing clinician effort.

INTRODUCTION: The anterior medial temporal lobe (MTL), including the entorhinal cortex (ERC) and Brodmann area 35 (BA35), is among the earliest cortical sites of tau pathology in Alzheimer's disease (AD), yet conventional image segmentation methods poorly capture these regions.

METHODS: We applied an automated segmentation approach using an extended Automatic Segmentation of Hippocampal Subfields (ASHS) atlas, including anterior MTL subregions, in 448 Pennsylvania Alzheimer's Disease Research Center participants with magnetic resonance imaging, tau positron emission tomography (PET) (n = 199), and/or plasma phosphorylated tau 217 (p-tau217) (n = 377). Amyloid beta (Aβ) positivity was defined using PET or plasma.

RESULTS: Tau-PET showed an anterior-posterior gradient, with highest uptake in BA35, ERC, and anterior hippocampus. Increased MTL tau-PET uptake and plasma p-tau217 were associated with cortical thinning localized to BA35 and ERC, even in cognitively unimpaired Aβ-positive individuals.

CONCLUSIONS: Anterior MTL subregions, especially BA35, show early vulnerability to tau-related neurodegeneration. Extended anterior MTL parcellation improves localization of early tau-associated structural changes and may facilitate biological staging in preclinical AD.