The Impact of Race, Ethnicity, and Sex on Fairness in Artificial Intelligence for Glaucoma Prediction Models.

Objective: Despite advances in artificial intelligence (AI) in glaucoma prediction, most works lack multicenter focus and do not consider fairness concerning sex, race, or ethnicity. This study aims to examine the impact of these sensitive attributes on developing fair AI models that predict glaucoma progression to necessitating incisional glaucoma surgery. Design: Database study. Participants: Thirty-nine thousand ninety patients with glaucoma, as identified by International Classification of Disease codes from 7 academic eye centers participating in the Sight OUtcomes Research Collaborative. Methods: We developed XGBoost models using 3 approaches: (1) excluding sensitive attributes as input features, (2) including them explicitly as input features, and (3) training separate models for each group. Model input features included demographic details, diagnosis codes, medications, and clinical information (intraocular pressure, visual acuity, etc.), from electronic health records. The models were trained on patients from 5 sites (N = 27 999) and evaluated on a held-out internal test set (N = 3499) and 2 external test sets consisting of N = 1550 and N = 2542 patients. Main Outcomes and Measures: Area under the receiver operating characteristic curve (AUROC) and equalized odds on the test set and external sites. Results: Six thousand six hundred eighty-two (17.1%) of 39 090 patients underwent glaucoma surgery with a mean age of 70.1 (standard deviation 14.6) years, 54.5% female, 62.3% White, 22.1% Black, and 4.7% Latinx/Hispanic. We found that not including the sensitive attributes led to better classification performance (AUROC: 0.77-0.82) but worsened fairness when evaluated on the internal test set. However, on external test sites, the opposite was true: including sensitive attributes resulted in better classification performance (AUROC: external #1 - [0.73-0.81], external #2 - [0.67-0.70]), but varying degrees of fairness for sex and race as measured by equalized odds. Conclusions: Artificial intelligence models predicting whether patients with glaucoma progress to surgery demonstrated bias with respect to sex, race, and ethnicity. The effect of sensitive attribute inclusion and exclusion on fairness and performance varied based on internal versus external test sets. Prior to deployment, AI models should be evaluated for fairness on the target population. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Referral Sources Across Racial and Ethnic Groups at Alzheimer's Disease Research Centers.

Background: Despite the need to increase engagement of underrepresented groups (URG) in Alzheimer's disease and related dementias (ADRD) studies, enrollment remains low. Objective: Compare referral sources across racial and ethnic groups among participants enrolled in ADRC studies. Methods: Data for this cross-sectional secondary analysis were extracted from the National Alzheimer's Coordinating Center Uniform Data Set. We performed mixed effects logistic regression models using generalized estimating equations for professional referral versus non-professional referral by racial and ethnic group, adjusted for age, gender, education, visit year, and Clinical Dementia Rating scale (CDR) with a random effect for study site. Results: Included in the analysis were 48,330 participants across 46 ADRCs (mean [SD] age, 71.3 [10.5] years; 20,767 female [57%]; 4,138 Hispanic [8.6%]; 1,392 non-Hispanic Asian [2.9%]; 6,766 non-Hispanic Black [14%] individuals; and 676 individuals [1.4%] of other races. Non-Hispanic Black and Asian participants had lower odds of being referred by a professional contact compared to non-Hispanic White participants (Black: adjusted OR = 0.61, 95% CI = 0.44-0.86, p = 0.005; Asian: adjusted OR = 0.65, 95% CI, p = 0.004). In participants who had completed an MRI, there was no significant difference in referral source across ethnic and racial groups. Conclusions: Further studies are needed to better understand the systemic and structural factors that contribute to differences in referral sources and disparities in recruitment of URG into ADRD studies.

Modeling primary microcephaly with human brain organoids reveals fundamental roles of CIT kinase activity.

Brain size and cellular heterogeneity are tightly regulated by species-specific proliferation and differentiation of multipotent neural progenitor cells (NPCs). Errors in this process are among the mechanisms of primary hereditary microcephaly (MCPH), a group of disorders characterized by reduced brain size and intellectual disability. Biallelic CIT missense variants that disrupt kinase function (CITKI/KI) and frameshift loss-of-function variants (CITFS/FS) are the genetic basis for MCPH17; however, the function of CIT catalytic activity in brain development and NPC cytokinesis is unknown. Therefore, we created the CitKI/KI mouse model and found that it does not phenocopy human microcephaly, unlike biallelic CitFS/FS animals. Nevertheless, both Cit models exhibited binucleation, DNA damage, and apoptosis. To investigate human-specific mechanisms of CIT microcephaly, we generated CITKI/KI and CITFS/FS human forebrain organoids. We found that CITKI/KI and CITFS/FS organoids lose cytoarchitectural complexity, transitioning from pseudostratified to simple neuroepithelium. This change was associated with defects that disrupt polarity of NPC cytokinesis, in addition to elevating apoptosis. Together, our results indicate that both CIT catalytic and scaffolding functions in NPC cytokinesis are critical for human corticogenesis. Species differences in corticogenesis and the dynamic 3D features of NPC mitosis underscore the utility of human forebrain organoid models for understanding human microcephaly.

Refractory Epilepsy in Adult Patient With COQ8A Variant Improves With CoQ10 Supplementation: A Case for Exome Sequencing in the ICU.

Objectives: Describe a case of stroke-like episodes and refractory status epilepticus diagnosed with primary CoQ10 deficiency-4 (COQ10D4) using whole-exome sequencing in the intensive care unit (ICU), with treatment implications. Methods: A patient presented to the emergency department with 1 month of progressively worsening focal motor status epilepticus and stroke-like imaging abnormalities. Multiple seizure medications, ketogenic diet, and elective intubation for anesthetic drips failed to achieve sustained seizure freedom. Genetic testing was pursued for prognostic information and identified potential treatment. Results: Whole-exome sequencing revealed compound heterozygous variants of COQ8A, including 1 allele not previously described as pathogenic. The patient's history, imaging, and genetic testing supported a diagnosis of COQ10D4. High-dose coenzyme Q10 supplementation was started with gradual clinical improvement. Discussion: Whole-exome sequencing is a fast and cost-effective means to diagnose rare neurologic disease in critically ill patients and can uncover treatment options. While primarily used in the neonatal ICU, appropriately selected adult patients may also benefit.

Beyond PhacoTrainer: Deep Learning for Enhanced Trabecular Meshwork Detection in MIGS Videos.

Purpose: The purpose of this study was to develop deep learning models for surgical video analysis, capable of identifying minimally invasive glaucoma surgery (MIGS) and locating the trabecular meshwork (TM). Methods: For classification of surgical steps, we had 313 video files (265 for cataract surgery and 48 for MIGS procedures), and for TM segmentation, we had 1743 frames (1110 for TM and 633 for no TM). We used transfer learning to update a classification model pretrained to recognize standard cataract surgical steps, enabling it to also identify MIGS procedures. For TM localization, we developed three different models: U-Net, Y-Net, and Cascaded. Segmentation accuracy for TM was measured by calculating the average pixel error between the predicted and ground truth TM locations. Results: Using transfer learning, we developed a model which achieved 87% accuracy for MIGS frame classification, with area under the receiver operating characteristic curve (AUROC) of 0.99. This model maintained a 79% accuracy for identifying 14 standard cataract surgery steps. The overall micro-averaged AUROC was 0.98. The U-Net model excelled in TM segmentation with an Intersection over union (IoU) score of 0.9988 and an average pixel error of 1.47. Conclusions: Building on prior work developing computer vision models for cataract surgical video, we developed models that recognize MIGS procedures and precisely localize the TM with superior performance. Our work demonstrates the potential of transfer learning for extending our computer vision models to new surgeries without the need for extensive additional data collection. Translational Relevance: Computer vision models in surgical videos can underpin the development of systems offering automated feedback for trainees, improving surgical training and patient care.

Associations between per- and poly-fluoroalkyl substance (PFAS) exposure and immune responses among women in the California Teachers study: A cross-sectional evaluation.

INTRODUCTION: Per- and polyfluoroalkyl substances (PFAS) are persistent environmental contaminants that have been linked to a number of health outcomes, including those related to immune dysfunction. However, there are limited numbers of epidemiological-based studies that directly examine the association between PFAS exposure and immune responses. METHODS: In this cross-sectional study nested in the California Teachers Study cohort, we measured nine PFAS analytes in serum. Of the 9 analytes, we further evaluated four (PFHxS [perfluorohexane sulfonate], PFNA [perfluorononanoic acid], PFOA [perfluorooctanoic acid], PFOS [perfluorooctanesulfonic acid]) that had detection levels of > 80 %, in relation to 16 systemic inflammatory/immune markers and corresponding immune pathways (Th1 [pro-inflammatory/macrophage activation], B-cell activation, and T-cell activation). Study participants (n = 722) were female, completed a questionnaire regarding various health measures and behaviors, and donated a blood sample between 2013-2016. The association between PFAS analytes and individual immune markers and pathways were evaluated by calculating odds ratios (OR) and 95 % confidence intervals (CI) in a logistic regression model. PFAS analytes were evaluated both as a dichotomous exposure (above or below the respective median) and as a continuous variable (per 1 unit increase [ng/mL]). RESULTS: The prevalence of detecting any PFAS analyte rose with increasing age, with the highest PFAS prevalence observed among those aged 75 + years and the lowest PFAS prevalence observed among those aged 40-49 years (study participant age range: 40-95 years). Significant associations with BAFF (B-cell activating factor) levels above the median were observed among participants with elevated (defined as above the median) levels of PFHxS (OR=1.53), PFOA (OR=1.43), and PFOS (OR=1.40). Similarly, there were statistically significant associations between elevated levels of PFHxS and TNFRII (tumor necrosis factor receptor 2) levels (OR=1.78) and IL2Rα (interleukin 2 receptor subunit alpha) levels (OR=1.48). We also observed significant inverse associations between elevated PFNA and sCD14 (soluble cluster of differentiation 14) (OR=0.73). No significant associations were observed between elevated PFNA and any immune marker. Evaluation of PFAS exposures as continuous exposures in association with dichotomized cytokines were generally consistent with the dichotomized associations. CONCLUSIONS: PFAS exposure was associated with altered levels of circulating inflammatory/immune markers; the associations were specific to PFAS analyte and immune marker. If validated, our results may suggest potential immune mechanisms underlying associations between the different PFAS analytes and adverse health outcomes.

Gap Analysis of Glaucoma Examination Concept Representations within Standard Systemized Nomenclature of Medicine - Clinical Terms.

OBJECTIVE/PURPOSE: Standardization of eye care data is important for clinical interoperability and research. We aimed to address gaps in the representations of glaucoma examination concepts within Systemized Nomenclature of Medicine - Clinical Terms (SNOMED-CT), the preferred terminology of the American Academy of Ophthalmology. DESIGN: Study of data elements. METHODS: Structured eye examination data fields from 2 electronic health records (EHR) systems (Epic Systems and Medisoft) were compared against existing SNOMED-CT codes for concepts representing glaucoma examination findings. Glaucoma specialists from multiple institutions were surveyed to identify high-priority gaps in representation, which were discussed among the SNOMED International Eye Care Clinical Reference Group. Proposals for new codes to address the gaps were formulated and submitted for inclusion in SNOMED-CT. MAIN OUTCOME MEASURES: Gaps in SNOMED-CT glaucoma examination concept representations. RESULTS: We identified several gaps in SNOMED-CT regarding glaucoma examination concepts. A survey of glaucoma specialists identified high-priority data elements within the categories of tonometry and gonioscopy. For tonometry, there was consensus that we need to define new codes related to maximum intraocular pressure (IOP) and target IOP and delineate all methods of measuring IOP. These new codes were proposed and successfully added to SNOMED-CT for future use. Regarding gonioscopy, the current terminology did not include the ability to denote the gonioscopic grading system used (e.g., Shaffer or Spaeth), degree of angle pigmentation, iris configuration (except for plateau iris), and iris approach. There was also no ability to specify eye laterality or angle quadrant for gonioscopic findings. We proposed a framework for representing gonioscopic findings as observable entities in SNOMED-CT. CONCLUSION: There are existing gaps in the standardized representation of findings related to tonometry and gonioscopy within SNOMED-CT. These are important areas for evaluating clinical outcomes and enabling secondary use of EHR data for glaucoma research. This international multi-institutional collaborative process enabled identification of gaps, prioritization, and development of data standards to address these gaps. Addressing these gaps and augmenting SNOMED-CT coverage of glaucoma examination findings could enhance clinical documentation and future research efforts related to glaucoma. FINANCIAL DISCLOSURE(S): Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Neuropsychiatric symptom burden in early-onset and late-onset Alzheimer's disease as a function of age.

INTRODUCTION: We examined the burden of neuropsychiatric symptoms (NPSs) in early-onset (EO) and late-onset (LO) Alzheimer's disease (AD) and adjusted for age effects via the inclusion of cognitively unimpaired (CU) individuals. METHODS: Cross-sectional data from 2940 EOAD, 8665 LOAD, and 8775 age-stratified CU individuals (early-CU, n = 2433; late-CU, n = 6342) from the National Alzheimer's Coordinating Center database were included. Fisher's exact tests compared EOAD and LOAD on the presence and severity of NPSs. Multiple logistic regression models included an age*diagnosis interaction to examine age effects. RESULTS: Presence (ps < 0.0001) and severity (ps < 0.05) of NPS were greater in EOAD than in LOAD. However, after adjusting for base rates in NPS in CU individuals (age effects), only elation and eating behaviors were more frequent in EOAD (ps < 0.05) and nighttime behaviors more frequent and severe in LOAD (ps < 0.05). DISCUSSION: Few NPSs were specific to the EOAD versus LOAD. Previous findings of greater NPS burden in EOAD may partially reflect age effects. HIGHLIGHTS: Adjusting for age effect, elation and eating problems are more frequent in EOAD. Adjusting for age effect, sleep disturbances are more frequent and severe in LOAD. Age effects underlie higher neuropsychiatric symptom presentation in EOAD than in LOAD.

Predicting Glaucoma Surgical Outcomes Using Neural Networks and Machine Learning on Electronic Health Records.

Purpose: To develop machine learning (ML) and deep learning (DL) models to predict glaucoma surgical outcomes, including postoperative intraocular pressure, use of ocular antihypertensive medications, and need for repeat surgery. Methods: We identified glaucoma surgeries performed at Stanford from 2013-2024, with two or more postoperative visits with intraocular pressure (IOP) measurement. Patient features were identified from the electronic health record (EHR), including demographics, prior diagnosis and procedure codes, medications and eye exam findings. Classical ML and DL models were developed to predict which glaucoma surgeries would result in surgical failure, defined as (1) IOP not reduced by more than 20% of preoperative baseline on two consecutive postoperative visits, (2) increased classes of glaucoma medications, and (3) need for additional glaucoma surgery or revision of original surgery. Results: A total of 2398 glaucoma surgeries of 1571 patients were included, of which 1677 surgeries met failure criteria. Random forest performed best for prediction of overall surgical failure, with accuracy of 75.5% and area under the receiver operator curve (AUROC) of 76.7%, similar to the deep learning model (accuracy 75.5%, AUROC 76.6%). Across all models, prediction performance was better for IOP outcomes (AUROC 86%) than need for an additional surgery (AUROC 76%) or need for additional glaucoma medication (AUC 70%). Conclusions: ML and DL algorithms can predict glaucoma surgery outcomes using structured data inputs from EHRs. Translational Relevance: Models that predict outcomes of glaucoma surgery may one day provide the basis for clinical decision support tools supporting surgeons in personalizing glaucoma treatment plans.

Hurricane Harvey and the risk of spontaneous preterm and early-term birth.

Background: Hurricane Harvey made landfall in August 2017 and resulted in catastrophic flooding in Houston, Texas. Prior studies of hurricanes and preterm birth have found conflicting results. We tested the hypotheses that exposure to Hurricane Harvey was associated with a higher risk of spontaneous pre- and early-term birth and assessed vulnerable subpopulations. Methods: We conducted a retrospective study of singleton births using administrative birth records in the nine-county greater Houston area from 2015 to 2019. We estimated the likelihood of pre- and early-term births using logistic regression, comparing births occurring during or within 1, 2, or 4 weeks of Hurricane Harvey to unexposed reference periods encompassing the same dates 2 years prior and after. Stratified models assessed effect modification by degree of flooding, birth parent age, high- vs. low-risk pregnancy, race/ethnicity, and prenatal care. Results: Among 15,564 births, we found no association between exposure to Hurricane Harvey and spontaneous preterm birth within 1 week adjusted (odds ratio [OR], 1.06; 95% confidence interval [CI] = 0.91, 1.25) but a 14% higher odds of spontaneous early-term birth (OR, 1.14; 95% CI = 1.04, 1.25). The odds of early-term birth were even higher in neighborhoods with severe flooding (OR, 1.21; 95% CI = 1.05, 1.38), segregated neighborhoods (OR, 1.23; 95% CI = 1.03, 1.47), and among foreign-born Hispanics (OR, 1.21; 95% CI = 1.04, 1.53) and pregnant people receiving no prenatal care (OR, 1.37; 95% CI = 1.03, 1.82). Effect estimates were attenuated or null when considering 2-week or 4-week lags to define exposure. Conclusions: Hurricane Harvey was associated with higher odds of spontaneous early-term birth up to 1 week later, especially among socially marginalized populations.

Black Americans' perceptions of Alzheimer's disease, a healthy brain, and strategies for brain health promotion.

Objectives: Lack of awareness of Alzheimer's disease (AD) among Black Americans may undermine their ability to identify potential AD risk. We examined Black Americans' perceptions and knowledge of AD, and views of a healthy brain, which may contribute to the development of effective and culturally sensitive strategies to address racial disparities in AD. Methods: We conducted a mixed-methods study, integrating a cross-sectional survey of 258 older (>55 years) Black participants and qualitative interviews with a sub-sample of N = 29. Both data sets were integrated to inform the results. Results: Participants endorsed having little knowledge of AD. While most participants reported practicing a healthy lifestyle to promote a healthy brain, the range of activities listed were limited. Participants made several suggestions to increase AD awareness, which includes using AD educational materials containing information that would benefit the whole family, not only older adults. Outreach approaches that address both individual behaviors and structural factors were also encouraged. Conclusion: Our findings identify ongoing needs to improve AD awareness among traditionally under-represented groups. Innovation: The study utilized novel approaches to examine participants' perspectives of AD that included a diverse sample of research naïve participants, and integrated exploration of participants' views of AD and brain health.

Reference values for leg effort during incremental cycle ergometry in non-trained healthy men and women, aged 19-85.

Heightened sensation of leg effort contributes importantly to poor exercise tolerance in patient populations. We aim to provide a sex- and age-adjusted frame of reference to judge symptom's normalcy across progressively higher exercise intensities during incremental exercise. Two-hundred and seventy-five non-trained subjects (130 men) aged 19-85 prospectively underwent incremental cycle ergometry. After establishing centiles-based norms for Borg leg effort scores (0-10 category-ratio scale) versus work rate, exponential loss function identified the centile that best quantified the symptom's severity individually. Peak O2 uptake and work rate (% predicted) were used to threshold gradually higher symptom intensity categories. Leg effort-work rate increased as a function of age; women typically reported higher scores at a given age, particularly in the younger groups (p < 0.05). For instance, "heavy" (5) scores at the 95th centile were reported at ~200 W (<40 years) and ~90 W (≥70 years) in men versus ~130 W and ~70 W in women, respectively. The following categories of leg effort severity were associated with progressively lower exercise capacity: ≤50th ("mild"), >50th to <75th ("moderate"), ≥75th to <95th ("severe"), and ≥ 95th ("very severe") (p < 0.05). Although most subjects reporting peak scores <5 were in "mild" range, higher scores were not predictive of the other categories (p > 0.05). This novel frame of reference for 0-10 Borg leg effort, which considers its cumulative burden across increasingly higher exercise intensities, might prove valuable to judging symptom's normalcy, quantifying its severity, and assessing the effects of interventions in clinical populations.

Exploratory pilot study to characterize the immune landscapes of malignant pleural effusions and their corresponding primary tumors from patients with breast carcinoma and lung adenocarcinoma.

INTRODUCTION: Malignant pleural effusion (MPE) is a frequent complication of advanced malignancies. In this pilot study, we characterized the immune landscapes of MPEs, compared them to their primary tumor (PT) samples from breast carcinoma (BC) and lung adenocarcinoma (LADC), and tested the utility of multiplexed image technology in cytological samples. MATERIALS AND METHODS: We evaluated the immune contexture of 6 BC and 5 LADC MPEs and their PTs using 3 multiplex immunofluorescence panels. We explored the associations between sample characteristics and pleural effusion-free survival. RESULTS: No MPE samples had positive programmed death-ligand 1 expression in malignant cells, although 3 of 11 PTs has positive programmed death-ligand 1 expression (more than 1% expression in malignant cells). Overall, in LADC samples, cluster of differentiation 3 (CD3)+ T cells and CD3+CD8+ cytotoxic T cells predominated (median percentages for MPEs versus PTs: 45.6% versus 40.7% and 4.7% versus 6.6%, respectively) compared with BC. CD68+ macrophages predominated in the BC samples (medians for MPEs 61.2% versus PTs for 57.1%) but not in the LADC samples. Generally in PTs, CD3+CD8+ forkhead box P3+ T cells and the median distances from the malignant cells to CD3+CD8+Ki67+ and CD3+ programmed cell death protein 1 + T cells correlated to earlier MPE after PT diagnosis. CONCLUSIONS: The immune cell phenotypes in the MPEs and PTs were similar within each cancer type but different between BC versus LADC. An MPE analysis can potentially be used as a substitute for a PT analysis, but an expanded study of this topic is essential.

Validation of Artificial Intelligence (AI)-Assisted Flow Cytometry Analysis for Immunological Disorders.

Flow cytometry is a vital diagnostic tool for hematologic and immunologic disorders, but manual analysis is prone to variation and time-consuming. Over the last decade, artificial intelligence (AI) has advanced significantly. In this study, we developed and validated an AI-assisted flow cytometry workflow using 379 clinical cases from 2021, employing a 3-tube, 10-color flow panel with 21 antibodies for primary immunodeficiency diseases and related immunological disorders. The AI software (DeepFlow™, version 2.1.1) is fully automated, reducing analysis time to under 5 min per case. It interacts with hematopatholoists for manual gating adjustments when necessary. Using proprietary multidimensional density-phenotype coupling algorithm, the AI model accurately classifies and enumerates T, B, and NK cells, along with important immune cell subsets, including CD4+ helper T cells, CD8+ cytotoxic T cells, CD3+/CD4-/CD8- double-negative T cells, and class-switched or non-switched B cells. Compared to manual analysis with hematopathologist-determined lymphocyte subset percentages as the gold standard, the AI model exhibited a strong correlation (r > 0.9) across lymphocyte subsets. This study highlights the accuracy and efficiency of AI-assisted flow cytometry in diagnosing immunological disorders in a clinical setting, providing a transformative approach within a concise timeframe.

Prediction Models for Glaucoma in a Multicenter Electronic Health Records Consortium: The Sight Outcomes Research Collaborative.

Purpose: Advances in artificial intelligence have enabled the development of predictive models for glaucoma. However, most work is single-center and uncertainty exists regarding the generalizability of such models. The purpose of this study was to build and evaluate machine learning (ML) approaches to predict glaucoma progression requiring surgery using data from a large multicenter consortium of electronic health records (EHR). Design: Cohort study. Participants: Thirty-six thousand five hundred forty-eight patients with glaucoma, as identified by International Classification of Diseases (ICD) codes from 6 academic eye centers participating in the Sight OUtcomes Research Collaborative (SOURCE). Methods: We developed ML models to predict whether patients with glaucoma would progress to glaucoma surgery in the coming year (identified by Current Procedural Terminology codes) using the following modeling approaches: (1) penalized logistic regression (lasso, ridge, and elastic net); (2) tree-based models (random forest, gradient boosted machines, and XGBoost), and (3) deep learning models. Model input features included demographics, diagnosis codes, medications, and clinical information (intraocular pressure, visual acuity, refractive status, and central corneal thickness) available from structured EHR data. One site was reserved as an "external site" test set (N = 1550); of the patients from the remaining sites, 10% each were randomly selected to be in development and test sets, with the remaining 27 999 reserved for model training. Main Outcome Measures: Evaluation metrics included area under the receiver operating characteristic curve (AUROC) on the test set and the external site. Results: Six thousand nineteen (16.5%) of 36 548 patients underwent glaucoma surgery. Overall, the AUROC ranged from 0.735 to 0.771 on the random test set and from 0.706 to 0.754 on the external test site, with the XGBoost and random forest model performing best, respectively. There was greatest performance decrease from the random test set to the external test site for the penalized regression models. Conclusions: Machine learning models developed using structured EHR data can reasonably predict whether glaucoma patients will need surgery, with reasonable generalizability to an external site. Additional research is needed to investigate the impact of protected class characteristics such as race or gender on model performance and fairness. Financial Disclosures: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Mindful awareness and resilience skills for adolescents (MARS-A): a mixed-methods study of a mindfulness-based intervention for a heterogeneous adolescent clinical population.

OBJECTIVES: Mindful Awareness and Resilience Skills for Adolescents (MARS-A) is a mindfulness-based intervention adapted for the adolescent population. While previous studies have explored the benefits of MARS-A in various single-diagnosis populations, the aim of this study was to assess MARS-A for a heterogenous clinical adolescent population with mental health and/or chronic diagnoses, focusing on the underlying suffering present in all these conditions rather than its effects on a single diagnosis itself. METHODS: Qualitative data was collected through interviews to understand post-intervention participant perspectives and experiences. Quantitative data was collected through measures to investigate preliminary secondary outcomes. RESULTS: After participating in MARS-A, participants reported qualitative benefits in enhanced well-being, including coping with difficult emotions and managing sleep and/or pain. Quantitative results showed a reduction in functional disability, psychological distress, perceived stress, and depressive symptoms; increase in positive affect; and benefit in coping with pain and chronic conditions. CONCLUSIONS: MARS-A shows great potential in a heterogeneous clinical adolescent population.

Mobile Critical Care Recovery Program for Survivors of Acute Respiratory Failure: A Randomized Clinical Trial.

Importance: Over 50% of Acute Respiratory Failure (ARF) survivors experience cognitive, physical, and psychological impairments that negatively impact their quality of life (QOL). Objective: To evaluate the efficacy of a post-intensive care unit (ICU) program, the Mobile Critical Care Recovery Program (m-CCRP) consisting of a nurse care coordinator supported by an interdisciplinary team, in improving the QOL of ARF survivors. Design, Setting, and Participants: This randomized clinical trial with concealed outcome assessments among ARF survivors was conducted from March 1, 2017, to April 30, 2022, with a 12-month follow-up. Patients were admitted to the ICU services of 4 Indiana hospitals (1 community, 1 county, 2 academic), affiliated with the Indiana University School of Medicine. Intervention: A 12-month nurse-led collaborative care intervention (m-CCRP) supported by an interdisciplinary group of clinicians (2 intensivists, 1 geriatrician, 1 ICU nurse, and 1 neuropsychologist) was compared with a telephone-based control. The intervention comprised longitudinal symptom monitoring coupled with nurse-delivered care protocols targeting cognition, physical function, personal care, mobility, sleep disturbances, pain, depression, anxiety, agitation or aggression, delusions or hallucinations, stress and physical health, legal and financial needs, and medication adherence. Main Outcomes and Measures: The primary outcome was QOL as measured by the 36-item Medical Outcomes Study Short Form Health Survey (SF-36) physical component summary (PCS) and mental component summary (MCS), with scores on each component ranging from 0-100, and higher scores indicating better health status. Results: In an intention-to-treat analysis among 466 ARF survivors (mean [SD] age, 56.1 [14.4] years; 250 [53.6%] female; 233 assigned to each group), the m-CCRP intervention for 12 months did not significantly improve the QOL compared with the control group (estimated difference in change from baseline between m-CCRP and control group: 1.61 [95% CI, -1.06 to 4.29] for SF-36 PCS; -2.50 [95% CI, -5.29 to 0.30] for SF-36 MCS. Compared with the control group, the rates of hospitalization were higher in the m-CCRP group (117 [50.2%] vs 95 [40.8%]; P = .04), whereas the 12-month mortality rates were not statistically significantly lower (24 [10.3%] vs 38 [16.3%]; P = .05). Conclusions and Relevance: Findings from this randomized clinical trial indicated that a nurse-led 12-month comprehensive interdisciplinary care intervention did not significantly improve the QOL of ARF survivors after ICU hospitalization. These results suggest that further research is needed to identify specific patient groups who could benefit from tailored post-ICU interventions. Trial Registration: ClinicalTrials.gov Identifier: NCT03053245.

Using Natural Language Processing to Identify Different Lens Pathology in Electronic Health Records.

PURPOSE: Nearly all published ophthalmology-related Big Data studies rely exclusively on International Classification of Diseases (ICD) billing codes to identify patients with particular ocular conditions. However, inaccurate or nonspecific codes may be used. We assessed whether natural language processing (NLP), as an alternative approach, could more accurately identify lens pathology. DESIGN: Database study comparing the accuracy of NLP versus ICD billing codes to properly identify lens pathology. METHODS: We developed an NLP algorithm capable of searching free-text lens exam data in the electronic health record (EHR) to identify the type(s) of cataract present, cataract density, presence of intraocular lenses, and other lens pathology. We applied our algorithm to 17.5 million lens exam records in the Sight Outcomes Research Collaborative (SOURCE) repository. We selected 4314 unique lens-exam entries and asked 11 clinicians to assess whether all pathology present in the entries had been correctly identified in the NLP algorithm output. The algorithm's sensitivity at accurately identifying lens pathology was compared with that of the ICD codes. RESULTS: The NLP algorithm correctly identified all lens pathology present in 4104 of the 4314 lens-exam entries (95.1%). For less common lens pathology, algorithm findings were corroborated by reviewing clinicians for 100% of mentions of pseudoexfoliation material and 99.7% for phimosis, subluxation, and synechia. Sensitivity at identifying lens pathology was better for NLP (0.98 [0.96-0.99] than for billing codes (0.49 [0.46-0.53]). CONCLUSIONS: Our NLP algorithm identifies and classifies lens abnormalities routinely documented by eye-care professionals with high accuracy. Such algorithms will help researchers to properly identify and classify ocular pathology, broadening the scope of feasible research using real-world data.

Spatial transcriptomic patterns underlying amyloid-ß and tau pathology are associated with cognitive dysfunction in Alzheimer's disease.

Amyloid-ß (Aß) and tau proteins accumulate within distinct neuronal systems in Alzheimer's disease (AD). Although it is not clear why certain brain regions are more vulnerable to Aß and tau pathologies than others, gene expression may play a role. We study the association between brain-wide gene expression profiles and regional vulnerability to Aß (gene-to-Aß associations) and tau (gene-to-tau associations) pathologies by leveraging two large independent AD cohorts. We identify AD susceptibility genes and gene modules in a gene co-expression network with expression profiles specifically related to regional vulnerability to Aß and tau pathologies in AD. In addition, we identify distinct biochemical pathways associated with the gene-to-Aß and the gene-to-tau associations. These findings may explain the discordance between regional Aß and tau pathologies. Finally, we propose an analytic framework, linking the identified gene-to-pathology associations to cognitive dysfunction in AD at the individual level, suggesting potential clinical implication of the gene-to-pathology associations.