Nevin Manimala

Indian J Ophthalmol. 2026 May 1;74(5):690-697. doi: 10.4103/IJO.IJO_3152_25. Epub 2026 Apr 29.

ABSTRACT

Anti-vascular endothelial growth factor (anti-VEGF) therapy is the mainstay of management for diabetic macular edema (DME), but marked variability in response, high injection frequency, and cumulative treatment burden highlight the need for tools that can individualize treatment beyond protocol-driven regimens. Artificial intelligence (AI) offers a pathway toward more individualized risk stratification and prognostic support primarily by capturing statistical associations rather than biological mechanisms. Deep learning systems have achieved great accuracy in detecting diabetic retinopathy (DR) and DME. Several autonomous DR/DME screening solutions are in clinical use. Recent advances have applied supervised machine learning, convolutional neural networks, generative adversarial networks, and ensemble methods to multimodal data from fundus images, baseline and follow-up optical coherence tomography (OCT), along with clinical and biochemical data, to classify likely responders and non-responders. These models automatically quantify and track imaging biomarkers to accurately predict central subfield thickness and vision outcomes after loading doses, and estimate future injection burden. AI-driven decision-support tools analyze vast amounts of patient data, treatment histories, and integrate multimodal data, including fundus images, OCT images, and systemic data to provide recommendations for optimal treatment and follow-up, tailored to each individual profile. The AI systems can potentially generate individualized risk and response profiles that can support decisions on initiating therapy, choosing between agents, tailoring treat-and-extend intervals, and timing switches to steroids or combination strategies. However, issues of generalizability, transparency, workflow integration, and ethical deployment need to be systematically addressed. AI-enabled decision support for patient selection and treatment response prediction is poised to become an integral component of anti-VEGF therapy.

PMID:42060353 | DOI:10.4103/IJO.IJO_3152_25

Traffic Inj Prev. 2026 Apr 30:1-13. doi: 10.1080/15389588.2026.2650661. Online ahead of print.

ABSTRACT

OBJECTIVE: This study aims to investigate the impact of combined fog and nighttime conditions on drivers’ cognitive load during tunnel approach, as reflected through fixation behavior. Specifically, it examines how these compounded adverse conditions influence visual attention patterns, including fixation duration, frequency, and spatial dispersion.

METHODS: A real-world driving experiment was conducted with 30 licensed drivers on the Xinjin Expressway. Eye movement data were collected using a Dikablis Pro eye tracker across four environmental scenarios: clear-day, foggy-day, clear-night, and foggy-night. The analysis focused on the tunnel approach zone, defined as the 10-s travel distance preceding the tunnel portal. Dependent variables included fixation duration, fixation frequency, horizontal fixation deviation, and vertical fixation deviation. One-way ANOVA and Tukey HSD post-hoc tests were employed to compare these metrics across scenarios.

RESULTS: The results revealed systematic variations in fixation behavior with increasing environmental complexity. Fixation duration was longest under foggy-night conditions (689.82 ± 30.4 ms) and shortest under clear-day conditions (325.59 ± 34.52 ms). Fixation frequency decreased progressively, with the highest rate in clear-day conditions (2.85 ± 0.18 Hz) and the lowest in foggy-night conditions (1.55 ± 0.17 Hz). Horizontal fixation deviation was largest in clear-day conditions (18.93 ± 2.91°) and smallest in foggy-night conditions (6.08 ± 1.68°), indicating lateral gaze constriction. Conversely, vertical fixation deviation increased significantly under adverse conditions, peaking in foggy-night scenarios (26.21 ± 3.74°), suggesting compensatory vertical scanning. All pairwise comparisons between scenarios were statistically significant (p < 0.01).

CONCLUSIONS: The combined effects of fog and nighttime conditions significantly elevate drivers’ cognitive load during tunnel approaches, manifesting as prolonged information processing, reduced attentional shifting, lateral visual field narrowing, and compensatory vertical search. These findings confirm the sensitivity of fixation-based metrics as indicators of cognitive load under compounded environmental stressors. The study provides empirical evidence for developing context-aware safety interventions, such as optimized tunnel lighting, adaptive traffic management, and enhanced driver assistance systems, tailored to mitigate cognitive overload in high-risk driving scenarios.

PMID:42060340 | DOI:10.1080/15389588.2026.2650661

J Glob Health. 2026 Apr 30;16:04144. doi: 10.7189/jogh.16.04144.

ABSTRACT

BACKGROUND: While there was a global shift in social interaction and alcohol consumption during the COVID-19 pandemic, their associations with changes in crime rates remain underexplored. We aimed to examine the associations between crime rates and alcohol use within the context of pandemic-related social distancing.

METHODS: We calculated crime rates across crime categories from 2011 to 2022 using crime statistics from the Korean National Police Agency. We estimated two linear regression models with the crime rate as the dependent variable. The first model examined the association of movie attendance (a proxy for social distancing) and the unemployment rate with crime rates. The second model additionally included per capita alcohol consumption to determine how the association between social distancing and crime rates was attenuated when accounting for alcohol use.

RESULTS: As of 2022, 19% of total crimes involved offenders under the influence of alcohol, with particularly high proportions in murder (64%), traffic accidents (47%), arson (32%), violence (28%), and rape (20%). Overall crime rates and offences committed under the influence of alcohol, which had steadily declined from 2011, fell sharply during the COVID-19 pandemic. Both alcohol consumption and social interaction declined during the pandemic. While the rates of overall crime, violence, rape, traffic accidents, and arson were initially associated with social contact, these associations were no longer significant after adjusting for alcohol consumption; instead, strong positive associations with alcohol consumption were observed. The rate of murder was not significantly associated with social contact, but exhibited a significant association only with alcohol consumption.

CONCLUSIONS: The concurrent declines in crime and alcohol consumption, along with the attenuating effect of alcohol in the relationship between social distancing and crime, suggest that addressing social drinking environments may be an effective strategy for reducing crime rates.

PMID:42060338 | DOI:10.7189/jogh.16.04144

J Chem Inf Model. 2026 Apr 30. doi: 10.1021/acs.jcim.5c02937. Online ahead of print.

ABSTRACT

Human β defensin type 3 (hBD-3) is recognized as one of the most intriguing antimicrobial peptides (AMPs) that holds the promise of solving drug resistance issues. hBD-3 can function (disruption of membrane integrity) in high salt environments, where most other AMPs fail. However, its functional mechanism at the molecular level remains elusive. To characterize its structure and dynamics during membrane crossing, long-time (a total of 57.0 μs) all-atom molecular dynamics simulations were conducted on hBD-3 monomers and dimers in both wild-type and analog (in which all three disulfide bonds are broken) forms that are embedded in four types of lipid membranes. Trajectory analysis was carried out using a statistical method─conformational dynamics analysis to calculate contact matrices and then principal component analysis (PCA) and linear discriminant analysis (LDA), in order to discern structural changes upon various physical and chemical perturbations. The result shows that the major collective coordinate primarily distinguishes between the wild-type and analog forms of hBD-3. For the hBD-3 monomer, the analog undergoes significant structural loss due to the lack of stabilizing disulfide bonds; salt exerts a nearly consistent effect on the contact degrees of freedom of the protein, whereas changes in lipid membrane composition have an insignificant effect. For the hBD-3 dimer, no consistent relationship between structure and salt concentration is indicated, and variations in the chemical composition of model bacterial membranes have a limited effect on its dynamics. These results suggest that the wild-type and analog forms of hBD-3 may employ different mechanisms when crossing bacterial membranes. The effect of salt on hBD-3 dynamics can be mitigated by the high net charge density of the protein. Additionally, the hBD-3 dimer can distinguish between model Gram-positive and Gram-negative membranes, whereas the monomer cannot. Overall, these findings provide unique insights into the structure, dynamics, and membrane-disrupting mechanism of hBD-3.

PMID:42060321 | DOI:10.1021/acs.jcim.5c02937

J Nurs Manag. 2026;2026(1):e2150047. doi: 10.1155/jonm/2150047.

ABSTRACT

PURPOSE: This study aimed to evaluate the effectiveness of a nurse-led educational program designed to improve nurse managers’ awareness, knowledge, and attitudes regarding climate change and its health impacts.

BACKGROUND/INTRODUCTION: Climate change is one of the most urgent global public health challenges, jeopardizing key determinants of health such as air quality, access to safe drinking water, food security, and adequate housing. As frontline healthcare providers, nurses are uniquely positioned to identify environmental health risks and promote climate-resilient healthcare practices. Despite this critical role, evidence suggests that nurses’ awareness and preparedness for climate-related health threats are insufficient. Therefore, strengthening climate literacy among nurse leaders is essential to enhance adaptation capacity in health systems.

METHODS: This study used a pretest-posttest experimental design with hospital-level randomization and included 108 nurse managers working in two public hospitals in Istanbul. Participants were randomly assigned to intervention and control groups. The study was conducted between March and June 2025, with data collection carried out between April and May 2025. Data were collected using a Descriptive Information Form and the Climate Change Awareness Scale (CCAS). The intervention group received 90 min of face-to-face training, including theoretical content, case-based learning, and interactive assessment, whereas the control group received a 90-min lecture after data collection on climate change and its health impacts, followed by a brief question-and-answer session. Measurements were taken at baseline, immediately after the intervention, and 1 month later. Data were analyzed using Friedman and Wilcoxon signed-rank tests.

RESULTS/FINDINGS: The intervention group demonstrated significant improvements in total and subscale CCAS scores at the postintervention time point compared with baseline and the control group (p < 0.05). Although a slight decrease was observed at the 1-month follow-up, scores remained higher than pretest levels. In the control group, although small differences were observed in certain subscales in the between-group comparisons, no statistically significant within-group changes were observed between the pretest, posttest, and follow-up scores.

DISCUSSION: The findings suggest that structured and nurse-led climate training for nurse managers has the potential to strengthen climate-related awareness and preparedness capacity.

CONCLUSION: Even short-term training increases nurse managers’ awareness of climate change and health.

TRIAL REGISTRATION: ClinicalTrials.gov identifier: NCT06905548.

PMID:42060318 | DOI:10.1155/jonm/2150047

JAMA Netw Open. 2026 Apr 1;9(4):e269347. doi: 10.1001/jamanetworkopen.2026.9347.

ABSTRACT

IMPORTANCE: The use of audiovisual aids in the consent process may improve comprehension and willingness to participate in research. However, data supporting this hypothesis are lacking.

OBJECTIVE: To determine the impact of audiovisual augmentation of the consent process on willingness to participate in research and on consent comprehension.

DESIGN, SETTING, AND PARTICIPANTS: The Personalizing Cardiovascular Health: A Population Approach to Promoting Cardiovascular Disease Resistance and Resilience Among Individuals With Obesity (RESILIENCE) nonrandomized cohort study embedded this multiarm randomized clinical trial to test different participant consent modalities. RESILIENCE recruited participants between September 2019 and March 2022, with the analysis for this randomized substudy conducted from July to December 2025. No follow-up was undertaken for this study. This component of the study was conducted exclusively virtually and completed online within a single large US academic health system.

INTERVENTIONS: Potential participants were randomized for this trial in a 1:1:1:1 allocation to receive consent information via (1) text-only (2) text and physician-featured video, (3) text and patient-featured video, or (4) text and animated video.

MAIN OUTCOMES AND MEASURES: The primary outcome was provision of consent to participate in the cohort clinical study. The secondary outcome was consent comprehension, based on a score of at least 5 of 7 on a follow-up comprehension survey. Robust log-linear Poisson regression was used to determine the relative risks (RRs) of outcomes in comparison with the text-only arm. Analyses were conducted with the evaluable population.

RESULTS: Of 1535 participants (968 [63.1%] female; 658 [42.9%] ≥60 years of age; 380 randomized to text-only, 386 to text and physician-featured video, 383 to text and patient-featured video, and 386 to text and animated video) who began the consent process, 888 (57.9%) overall gave consent to participate in the study. Patient consent rates were similar across the 4 consent delivery strategies (text only [comparator], 221 of 380 [58.2%]; text and physician video, 241 of 386 [62.4%]; RR, 1.07 [96% CI, 0.95-1.21]; text and patient video, 203 of 383 [53.0%]; RR, 0.91 [96% CI, 0.8-1.04]; and text and animated video, 223 of 386 [57.8%]; RR, 0.99 [96% CI, 0.88-1.13]). Among 884 individuals who consented and completed the consent comprehension survey, the rate of comprehension of the consent process was high 86.4% (n = 764) and did not vary among the intervention delivery strategies (text only, 193 of 221 [87.3%]; text and physician video, 207 of 241 [85.9%]; text and patient video, 177 of 203 [87.2%]; and text and animated video, 187 of 223 [83.9%]). Compared with the text-only arm, the RRs of consent comprehension were not different for the text and physician-featured video arm (RR, 1.00; [95% CI, 0.93-1.07]), the text and patient-featured video arm (RR, 1.00 [95% CI, 0.93-1.07]) and the text and animated video arm (RR, 0.96 [95% CI, 0.89-1.04]).

CONCLUSIONS AND RELEVANCE: In this randomized clinical trial of virtual consent delivery, the rates of patient consent and consent comprehension were similar regardless of whether they were delivered by text only or text augmented by physician, patient, or animated videos. Further evaluation of these consent delivery processes across different study types and patient populations is suggested to optimize the recruitment of diverse, informed populations.

TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT04551872.

PMID:42060306 | DOI:10.1001/jamanetworkopen.2026.9347

JAMA Netw Open. 2026 Apr 1;9(4):e269923. doi: 10.1001/jamanetworkopen.2026.9923.

ABSTRACT

IMPORTANCE: Approximately 1% of newborns receive advanced neonatal resuscitation interventions (ANRIs; consisting of endotracheal intubation, chest compressions, and epinephrine administration), the likelihood of which increases with known clinical factors. Associations of socioeconomic status (SES) and health system factors with ANRI are unknown.

OBJECTIVE: To examine the associations among maternal SES, remoteness of maternal residence, birth-site level of service, clinical factors, and ANRIs.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study used an administrative dataset of all live births at 34 weeks’ gestation or later between January 1, 2000, and December 31, 2020, in Alberta, Canada, a jurisdiction with universal health care and provincially coordinated Neonatal Resuscitation Program training. Hospital births and home births attended by registered midwives were included. Statistical analyses were conducted from August 1 to September 30, 2025.

EXPOSURES: Maternal SES (classified according to Pampalon Material Deprivation Index quintiles), remoteness of residence (categorized using the Statistics Canada Remoteness Index), birth-site level of service, and clinical factors (including maternal, intrapartum, and fetal or neonatal characteristics).

MAIN OUTCOMES AND MEASURES: The primary outcome was any ANRI. Secondary outcomes were individual interventions (endotracheal intubation, chest compression, epinephrine administration). Multivariable logistic regression with stepwise variable selection was used to examine associations between SES and health system factors and the primary outcome.

RESULTS: A total of 966 475 live births at 34 weeks’ gestation or later were analyzed, of whom 1.0% received ANRI. Neonates had a mean (SD) gestational age of 38.9 (1.48) weeks and included 494 835 males (51.2%); mothers had a mean (SD) age of 29.4 (5.46) years. Compared with level 2 birth sites, level 3 birth sites had lower odds of ANRI (odds ratio [OR], 0.57; 95% CI, 0.53-0.61) and all interventions. Births at level 1A sites (OR, 2.53; 95% CI, 1.99-3.15) and home births (OR, 1.44; 95% CI, 1.18-1.74) had higher odds of ANRI and chest compressions (OR, 7.32 [95% CI, 5.61-9.40] and 4.12 [95% CI, 3.29-5.11], respectively). All level 1 births (level 1C: OR, 0.54; 95% CI, 0.49-0.60; level 1B: OR, 0.52; 95% CI, 0.45-0.59; and level 1A: OR, 0.4 95% CI, 0.20-0.71) and home births (OR, 0.27; 95% CI, 0.16-0.43) were associated with lower odds of endotracheal intubations. There were no associations for maternal SES and maternal residence remoteness. The most significant clinical factors associated with ANRI were maternal general anesthesia (OR, 4.89; 95% CI, 4.47-5.34), lower gestational age (highest OR at 34 weeks: 3.60; 95% CI, 3.11-4.15), cesarean delivery (OR, 1.80; 95% CI, 1.60-2.02), meconium-stained amniotic fluid (OR, 2.05; 95% CI, 1.94-2.17), and macrosomia (OR, 1.83; 95% CI, 1.55-2.16).

CONCLUSIONS AND RELEVANCE: In this cross-sectional study, maternal SES and maternal residence remoteness were not associated with increased odds of ANRI, and birth sites with a lower level of service were associated with higher odds of chest compressions but lower odds of endotracheal intubations. Clinicians should be prepared to resuscitate neonates exposed to general anesthesia.

PMID:42060305 | DOI:10.1001/jamanetworkopen.2026.9923

JAMA Netw Open. 2026 Apr 1;9(4):e269930. doi: 10.1001/jamanetworkopen.2026.9930.

ABSTRACT

IMPORTANCE: Delays in hospital discharge to skilled nursing facilities (SNFs) are widely viewed by hospital administrators and policymakers as a significant contributor to hospital crowding and adverse patient outcomes. However, the extent to which regional SNF capacity is associated with hospital length of stay (LOS) is not well understood.

OBJECTIVES: To construct empirically derived hospital-SNF markets using a network science approach and to examine whether variation in SNF capacity within these markets is associated with hospital LOS.

DESIGN, SETTING, AND PARTICIPANTS: This cross-sectional study used 2018 and 2019 Medicare fee-for-service claims linked with publicly available SNF data and included 3.34 million inpatient admissions discharged to SNFs from short-term hospitals in the US. Data were analyzed from October 1, 2024, to February 28, 2026.

EXPOSURES: Monthly mean SNF occupancy rates and nurse and therapy staffing levels within an empirically defined regional hospital-SNF market from a patient-sharing network analysis.

MAIN OUTCOMES AND MEASURES: The main outcome was hospital LOS. Linear regression models estimated associations between market-level SNF capacity measures and LOS, adjusting for patient characteristics, hospital-level postacute care demand, SNF organizational features, and time and hospital-SNF market fixed effects. Analyses were stratified by dual Medicare-Medicaid eligibility.

RESULTS: The study sample included 3.34 million Medicare inpatient admissions (mean [SD] age, 78.6 [11.3] years; 60.2% women) discharged to SNFs across 62 303 hospital-SNF dyads, and 421 empirically derived hospital-SNF markets. The mean (SD) hospital LOS was 6.9 (6.8) days (7.4 [7.8] days for dual-eligible patients and 6.7 [6.2] days for non-dual-eligible patients). Within markets, each additional nurse staffing hour per patient-day was associated with a 3.5% shorter LOS (95% CI, -5.5% to -1.4%); associations were also found among dual-eligible patients (-3.9% [95% CI, -6.7% to -0.9%) and non-dual-eligible patients (-3.2% [95% CI, -5.4% to -0.9%]). Differences in occupancy rates and therapy staffing within networks were not associated with differences in LOS after adjusting for patient, hospital, and network factors.

CONCLUSIONS AND RELEVANCE: Hospital-SNF markets defined by the strength of patient-sharing relationships provide a reproducible framework for examining postacute care capacity across payer populations. Within these markets, for the Medicare fee-for-service population, higher nurse staffing levels were associated with shorter hospital LOS. Strategies to reduce discharge delays and hospital crowding may require closer alignment between hospitals and highly connected SNFs to maintain adequate postacute care capacity.

PMID:42060304 | DOI:10.1001/jamanetworkopen.2026.9930

JAMA Netw Open. 2026 Apr 1;9(4):e269937. doi: 10.1001/jamanetworkopen.2026.9937.

ABSTRACT

IMPORTANCE: Identifying factors that explain differences in physical performance measures by race and region in the US and whether they are associated with changes over time could inform preventive strategies.

OBJECTIVE: To examine 10-year changes in physical performance across race and race-region among Black and White older adults and whether differences are explained by demographic, cardiovascular, socioeconomic, and cognitive factors.

DESIGN, SETTING, AND PARTICIPANTS: The prospective Atherosclerosis Risk in Communities (ARIC) cohort study recruited participants between 1987 and 1989. This study included participants from 4 US communities in North Carolina, Mississippi, Maryland, and Minnesota who underwent physical performance evaluations initiated in 2011 to 2013 (visit 5 examinations) and in 3 subsequent in-person examinations conducted over a 10-year period through 2022. Data were analyzed from January through May 2025.

EXPOSURE: Self-identified race and region.

MAIN OUTCOMES AND MEASURES: Physical performance was assessed using the Short Physical Performance Battery (SPPB; score range, 0-12, with higher scores indicating better performance). Multivariable generalized estimating equations estimated cross-sectional differences as well as differences in 10-year changes by race and by race and region, adjusting for demographic, cardiovascular, socioeconomic, and cognitive factors.

RESULTS: Among 5666 participants (mean [SD] age, 75.4 [5.1] years; 3258 [58%] women), 1233 (22%) self-identified their race as Black and 4433 (78%) self-identified their race as White. Adjusting for demographic, cardiovascular, socioeconomic, and cognitive factors explained observed cross-sectional differences in physical performance between Black participants and White participants. However, 10-year SPPB longitudinal declines were steeper for Black participants compared with White participants (absolute difference, -0.80 points; 95% CI, -1.19 to -0.41 points), even after adjusting for all factors that accounted for cross-sectional differences; this finding met the 0.50-point threshold for a clinically meaningful difference. Between-region declines within races also showed notable differences; for example, SPPB declines were steeper for White Maryland participants compared with White North Carolina participants (absolute difference, -1.14 points; 95% CI, -1.45 to -0.83 points), and declines were steeper for Black Mississippi participants compared with Black North Carolina participants (absolute difference, -1.72 points; 95% CI, -2.65 to -0.78 points).

CONCLUSIONS AND RELEVANCE: In this cohort study of older adults, meaningful racial differences in physical performance declines were not explained by commonly suggested drivers of performance differences. Other unmeasured regional features may explain racial differences in decline in physical performance and warrant investigation.

PMID:42060303 | DOI:10.1001/jamanetworkopen.2026.9937

JAMA Netw Open. 2026 Apr 1;9(4):e269948. doi: 10.1001/jamanetworkopen.2026.9948.

ABSTRACT

IMPORTANCE: Children who survive critical illness commonly experience long-term morbidities. Little is known about the association of critical illness with cognitive health due to lack of preillness and postillness assessments and an adequate comparison population.

OBJECTIVE: To use school-based testing to evaluate cognitive health outcomes among children treated in the pediatric intensive care unit (PICU) compared with non-PICU-exposed control students.

DESIGN, SETTING, AND PARTICIPANTS: This retrospective case-control study used statewide academic data and propensity score matching of 1088 patients admitted to the only PICU in Arkansas from January 1, 2008, to December 31, 2018, as well as controls matched on sociodemographic and pre-PICU admission academic factors. Statistical analysis was performed from March 2024 to September 2025.

EXPOSURE: PICU admission.

MAIN OUTCOMES AND MEASURES: Primary outcomes were (1) return to standardized testing, assessed using multivariable logistic regression to assess the odds of having a standardized test within 2 years after PICU admission compared with controls, and (2) change in pre-PICU to post-PICU test scores for PICU patients compared with propensity-matched control students, assessed using multivariable linear regression.

RESULTS: In this case-control study of 1088 school-aged patients (mean [SD] age, 12.1 [1.6] years; 566 girls [52.0%]), fewer PICU patients than controls had test scores after admission for math (80.6% [874 of 1085] vs 86.5% [938 of 1085]; adjusted odds ratio [AOR], 0.64 [95% CI, 0.51-0.81]) and reading (81.1% [877 of 1081] vs 87.1% [941 of 1081]; AOR, 0.64 [95% CI, 0.51-0.82]). PICU patients’ preadmission z scores were below average in math (z = -0.23 [95% CI, -0.29 to -0.16]) and reading (z = -0.22 [95% CI, -0.29 to -0.15]) compared with Arkansas students in the same grade and year. In adjusted pre-post analyses, PICU patients had a small but significant decrease in reading relative to controls (-0.07 [95% CI, -0.14 to -0.01]). The change in math score was not statistically significant (-0.06 [95% CI, -0.13 to 0.003]).

CONCLUSIONS AND RELEVANCE: This study suggests that PICU patients were less likely to take standardized tests after discharge and that those who did had greater decreases in reading scores relative to matched controls. Future studies should identify risk factors for nonreturn to testing and score decrease.

PMID:42060302 | DOI:10.1001/jamanetworkopen.2026.9948