Browse Articles

Abstract TU212: Prediction of Early-Onset Preeclampsia Using Deep Learning: A Systematic Review of Clinical and Imaging Models

Background: Early-onset preeclampsia is a leading cause of maternal and perinatal morbidity and mortality. Deep learning (DL) offers a promising approach for early prediction, but a systematic assessment of its performance is needed. Objective: This systematic review aims to synthesize evidence from DL models for predicting early-onset preeclampsia using clinical and imaging data. Methods: We conducted a systematic review following the PRISMA 2020 guidelines. A comprehensive search of five electronic databases (PubMed, Embase, LILACS, Scopus, and Web of Science) was performed on June 11, 2025. Before study selection and data extraction, two reviewers, trained in a pilot session, conducted independent and blinded reviews. The risk of bias was assessed using the PROBAST tool. Data on study design, population, model type, input features, validation strategy, and performance metrics were extracted. Descriptive statistics and percentages were calculated to summarize key characteristics. Results: From a total of 15 included studies, sample sizes ranged from 100 to 360,943 participants. Descriptive analyses showed that 53.3% (n=8) of studies used DL exclusively, while 46.7% (n=7) combined DL with traditional machine learning. Clinical registries were the primary data source (93.3%, n=14). The majority of models (73.3%, n=11) integrated maternal characteristics, mean arterial pressure, and biomarkers such as PlGF. Imaging data, including electrocardiogram and retinal fundus images, were utilized in only 26.7% (n=4) of studies. Internal validation alone was reported in 80.0% (n=12) of studies, while both internal and external validation were performed in 20.0% (n=3). Model performance varied, with area under the curve (AUC) values ranging from 0.57 to 0.98. The highest performance (AUC 0.98) was achieved by a model using electrocardiogram data. Risk of bias assessment indicated a low overall risk for 80.0% (n=12) of studies, with the analysis domain being the most frequent source of high risk. Conclusions: DL models demonstrate significant potential for the early prediction of early-onset preeclampsia. However, limited external validation, variability in performance, and underutilization of imaging data underscore the necessity for standardized, prospective validation in diverse cohorts prior to clinical implementation.

Antibacterial and biocompatibility potentials of zinc oxide quantum dots via Nd: YAG laser ablation

Abstract MPTH66: No Significant Six-Month Changes in Lifestyle Behaviors Among Adults with Elevated BP or Stage 1 Hypertension in the LINKED-BP Program

Background: Lifestyle modification is central to hypertension prevention and control. Despite its recognized importance, few community-based clinical trials have evaluated short-term behavioral change within multicomponent interventions. Objective: To examine six-month changes in diet, sleep, and physical activity, among adults in a hypertension prevention trial [Home Blood Pressure Telemonitoring LINKED with CHWs to Improve Blood Pressure(LINKED-BP) Program (NCT05719647)] randomized to a multilevel intervention versus enhanced usual care. We hypothesized that participants in the intervention arm would have greater improvements in these lifestyle behaviors than those in the control arm. Methods: Participants were adults aged ≥18 years at risk for stage 2 hypertension who were enrolled in the LINKED-BP Program. This cluster randomized clinical trial was conducted across health systems in Maryland and the Washington, DC metropolitan area. Participants were randomized to a multi-level intervention with community health workers providing education and lifestyle modification support or enhanced usual care over a 12-month period. For this analysis, we included participants with baseline and 6-month survey data. We assessed six-month changes in diet (Mediterranean Eating Pattern for Americans score), nightly sleep duration (hours), and physical activity (International Physical Activity Questionnaire). Descriptive statistics and multivariable linear regression were used to analyze the association between study arm and changes in lifestyle behaviors. Results: Participants (N=335) had a mean age of 43.1 (± 12.2) years and 69% were female ( Table 1 ). Between baseline and 6 months, physical activity increased modestly across both groups (602 to 834 MET-min/week overall), but the change was not statistically significant (Table 2 ) . Other lifestyle behaviors remained largely unchanged. Multi-variable adjusted models comparing the intervention with enhanced usual care showed no significant between-group differences in the changes in diet (0.28 [–0.20, 0.75]), sleep (–0.03 [–0.30, 0.25]), or physical activity (51.4 [–206.9, 309.7]) (Table 3 ). Conclusions: Preliminary data from this cluster-randomized trial showed non-significant within-group improvements in physical activity over 6 months. Achieving meaningful behavioral change in real-world hypertension management may require more intensive, sustained, and equity-focused interventions.

Abstract TH802: Adverse Events Associated with Immune Checkpoint Inhibitors in a Real-World Hispanic Cohort: Analysis of Cardiovascular Outcomes

Background: Immune checkpoint inhibitors (ICIs) have transformed cancer therapy, yet data on cardiovascular and thrombotic safety in Hispanic populations are limited. Evidence from the Hispanic Community Health Study (HCHS) shows disproportionate cardiometabolic risk for major adverse cardiovascular events (MACE) and thrombosis, likely influenced by social determinants. Objective: We aimed to describe the incidence and predictors of MACE and thrombosis in ICI-treated adults at a Colombian quaternary center. Methods: We conducted an observational retrospective study including adults (≥18 y) treated with PD-1, PD-L1, or PD-1 + CTLA-4 inhibitors (2020–2024). Data were extracted from electronic records into REDCap. Outcomes were MACE, thrombosis, and irAEs. Event-free survival was estimated by Kaplan–Meier; predictors were assessed using multivariable Cox regression. Results: A total of 206 patients were included (median age 66 years, 59.7 % male). Most received PD-1 inhibitors (67.5 %), followed by PD-L1 (19.9 %) and combined PD-1 + CTLA-4 therapy (12.6 %). Overall, MACE occurred in 21 patients (10.2%), mainly myocardial infarction and heart failure, while thrombotic events were reported in 29 (14.1%), predominantly venous (11.7%). Median times to MACE and thrombosis were 302 and 274 days, respectively (Figure 1 and 2). The combination group showed the highest incidence of both complications (MACE 19.2%, thrombosis 26.9%) compared with PD-1 (9.4%, 14.4%) and PD-L1 (77.3% 4.9%), reaching statistical significance for thrombosis (p = 0.04). Concomitant chemotherapy independently increased thrombotic risk (HR 2.49, 95% CI 1.12–5.54, p = 0.025), whereas no covariates predicted MACE. Patients developing both MACE and thrombosis exhibited markedly reduced survival (HR 2.26, 95% CI 0.97–5.23, p = 0.058). Co-occurrence of these events was more frequent among high ASCVD-risk patients (30 %) than in intermediate (3.1 %) and low-risk (5.7 %) groups (p = 0.018), underscoring the prognostic value of baseline cardiovascular risk in ICI-treated populations. Conclusion: In this real-world Hispanic cohort, ICIs were associated with notable thrombotic and cardiovascular toxicity, especially with PD-1 + CTLA-4 therapy. The coexistence of MACE and thrombosis as well as elevated baseline ASCVD risk, identified patients with poorer survival, underscoring the need for cardio-oncologic surveillance and preventive strategies as immunotherapy expands in Latin America.

Abstract MPTU21: Proactive Outreach May Mitigate Disengagement in Electronic Healthy Eating Interventions: A Qualitative Study

Background: The US population eats less than recommended amounts of fruit and vegetables (F&V). We conducted a randomized clinical trial to test whether a $160 monthly F&V subsidy, choice architecture (F&V shown first on the online store), and/or text message reminders of the subsidy balance increased F&V purchases in an online grocery store over 3 months. We conducted interviews with high and low spending participants to understand their experiences. Methods: Using data on participants’ subsidy expenditures, we recruited respondents from the top and bottom 30% of average monthly expenditures among those in the intervention arm for the three months they participated. Participants were invited to brief telephone interviews that asked about their experience in the program, benefits, challenges, and program impact. Interviews were conducted March - May 2025, recorded, transcribed, and blinded. Two researchers summarized transcript responses. After all data were summarized, responses were unblinded, analyzed, and compared by group. Results: We conducted 41 interviews with 23 high spenders and 18 low spenders (N=187). Respondents were primarily Black, female, and over 50 years of age. High spenders described the program as easy to use, motivating, and enjoyable. They liked the convenience of selecting eligible items, tracking balances, and having groceries delivered. The program enhanced access to F&V while fitting into existing routines and helping participants save time. They also appreciated features such as reminder texts and the $25 sign-up incentive. Some high spenders reported difficulties using the site but proactively reached out for support and found solutions. Low spenders struggled with adoption, often due to technical or payment barriers. Confusion around the site, digital subsidy, and competing payment methods (e.g., SNAP or debit cards) discouraged participation. These participants frequently expressed frustration and reported abandoning attempts to use the subsidy. They emphasized the need for greater communication and troubleshooting support from the study team. Conclusion: While many high spenders found the program motivating and easy to use, lower spenders were often discouraged by technological barriers such as confusion with payment systems. Interventions that include an electronic component should consider proactive outreach to help mitigate these challenges and support full engagement.

A machine learning-derived biological age model for liver grafts provides a superior assessment of aging compared to chronological age in transplantation

Abstract TU252: Hospital-level Disparities and Long-term Survival in Patients With STEMI: Evidence from the China Acute Myocardial Infarction Registry

Objective: We aimed to examine the association between hospital level and 5-year mortality among ST-segment elevation myocardial infarction (STEMI) patients in China, and identify factors explaining these differences. Methods: We analyzed data from the China Acute Myocardial Infarction (CAMI) registry, which enrolled 12,697 STEMI patients from 108 hospitals across 31 provinces in mainland China between January 2013 and September 2014, with follow-up through January 1, 2020. Patients were categorized according to initial admission at provincial-, prefecture-, or county-level hospitals. The primary outcome was 5-year all-cause mortality. Kaplan–Meier estimates and multivariable Cox proportional hazards models were used to assess mortality risk across hospital levels, with competing risk models applied for cardiovascular deaths. Prespecified subgroup analyses evaluated effect modification by demographic and clinical factors. In addition, a multiple mediation framework was employed to quantify the contribution of patient characteristics and treatment differences to hospital-level disparities in long-term survival. Results: Among 12,697 patients (mean age, 62.8 years; 75.7% male), the 5-year all-cause mortality was 19.8%, 25.3%, and 36.2% at provincial-level, prefecture-level, and county-level hospitals, respectively. After multi-variable adjustment, compared with patients treated at provincial-level hospitals, the adjusted HRs for all-cause mortality were 1.24 (95% CI, 1.11-1.39) at prefecture-level hospitals and 1.54 (95% CI, 1.34-1.78) at county-level hospitals (P for trend <.001). Similar trends were observed for cardiovascular mortality. Subgroup analyses indicated that the disparities were most pronounced among patients with overweight or obesity, history of prior heart failure, those with cardiogenic shock at admission, and those with onset-to-arrival time>12h. Mediation analysis suggested that primary PCI use, age, and socioeconomic factors contributed substantially to the observed differences in mortality. Conclusions: In this nationwide cohort of STEMI patients, initial treatment at lower-level hospitals was associated with significantly higher long-term mortality. These findings highlight the need to improve access to timely, evidence-based care and to strengthen regional referral systems and treatment capacity at lower-tier hospitals to reduce outcome disparities.

Abstract MPTH73: Gut microbiome, blood metabolome and incident hypertension in Hispanic Community Health Study/Study of Latinos

Background: While gut dysbiosis is linked to prevalent hypertension (HTN), microbial features associated with incident HTN and the mediating role of circulating metabolites remain poorly understood. Methods: We conducted a prospective analysis of 1,265 HTN-free adults with concurrently profiling of gut metagenomics (440 core species with prevalance >10% and relative abundance >0.01%) and serum metabolomics (767 named metabolites) from HCHS/SOL (mean age 53 yrs; 65.8% female; 358 incident HTN over a median of 7 yrs of follow-up). Cox proportional hazard models were used to identify gut microbial species and metabolites associated with incident HTN. Mediation analysis was used to quantify the extent to which metabolites explain the microbiome-HTN relationship. Results: We identified 18 bacterial species associated with incident HTN (FDR<0.25, Fig1.a), largely independent of baseline cardiometabolic traits (Fig1.b). Seven species, predominantly Prevotella spp. (e.g., P. nanceiensis and P. histicola ), were associated with higher HTN risk, whereas the other 11 species showed inverse associations, including three Akkermansia spp. , two Firmicutes spp. , and multiple Clostridiales spp. (e.g., Intestinimonas massiliensis ). We further identified 224 metabolites (FDR<0.05) that were associated with both at least one HTN-related species and incident HTN. These metabolites collectively mediated a substantial portion of the observed microbiota-HTN associations, with the proportion of mediated effect ranging from 9% for Parabacteroides merdae to 75.2% for Intestinimonas massiliensis (Fig.1c). Key metabolites with significant mediating effects, especially on the protective association between Intestinimonas massiliensis and incident HTN, included microbially influenced steroids (e.g., androstenediol disulfate, tetrahydrocortisol glucuronide), N-acetyl-amino acids (e.g., N-acetylphenylalanine, N-acetyltyrosine), and markers of endothelial dysfunction and vascular stiffness (e.g., cysteine-glutathione disulfide, 5-hydroxylysine) (Fig1.d) . A combined panel of the identified gut microbiota and metabolites significantly improved HTN risk prediction beyond conventional risk factors (AUC improved from 0.77 to 0.84, p<0.001). Conclusion: Our study characterizes multiple bacterial species associated with incident HTN, and the associations might be partly mediated by circulating metabolites. Inclusion of gut microbiome and associated metabolites facilitates the risk prediction of HTN.

Abstract TU219: Impact of Adverse Pregnancy Outcomes on Long-term Outcomes in Women with Heart Failure

Introduction: Adverse pregnancy outcomes (APOs) are risk factors for future cardiovascular disease, including heart failure (HF). However, the impact of a history of APOs among women diagnosed with HF is unclear. We investigated whether women with HF and a history of APOs, including gestational hypertension, preeclampsia, eclampsia, gestational diabetes mellitus, and fetal growth restriction, experienced worse long-term HF-related outcomes than those without APOs. Methods: Using the TriNetX Global Collaborative Network, we conducted a retrospective cohort study of women aged 25-80 years with HF (both HFpEF and HFrEF) and a history of pregnancy. Cohort 1 included women with prior APOs (ICD O13-15, O16.1, O254.41, O36.59), and Cohort 2 included women with documented pregnancy but no APOs, all based on International Classification of Diseases, 10th Revision (ICD-10) encounter diagnosis. Cohorts were propensity-score matched (PSM) on 27 variables, including demographics and established cardiovascular comorbidities, yielding 5,061 matched pairs. Index event was the date of initial documented HF diagnosis (I50). Baseline characteristics were evaluated over 5 years preceeding the index date. Outcomes were assessed over 10 years following index and included HF exacerbations (I50.23, I50.33, I50.31, I50.21, I50.43, I50.51), ischemic stroke (I63), myocardial infarction (I21), and all-cause mortality. Statistical analyses were performed using Z-statistics and Kaplan-Meier survival analyses. Results: Cohorts were well matched demographically (mean age 45.7 vs 45.4 years; 47.9% vs 46.9% White; 36.7% vs 37.3% Black) and across all major cardiovascular risk factors after PSM. Women with APOs had significantly higher rates of HF exacerbations (44.6% vs 39.2%, HR 1.19, 95% CI 1.12–1.26, p<0.001). No significant differences were observed in ischemic stroke (p=0.284), myocardial infarction (p=0.166), or all-cause mortality (p=0.931). Conclusions: Among women with HF, a history of APOs is associated with more HF exacerbations. These findings highlight APO history as an important marker for identifying women who may benefit from more intensive HF management.

Investigation of carbonation-induced microstructural changes in low-cement concrete using sustainable binders: GGBS and calcium carbonate

Abstract Concrete manufacturing makes a major contribution to the world economy, especially the construction industry, but it has a high level of CO 2 emissions, which harm the environment. To address this environmental issue, this paper investigates the possibility of utilizing a natural mineral that consists of iron, silica, and magnesium, and that is referred to as calcium carbonate, to partially substitute cement. This has been neglected material in past studies, even though it can be used to boost the process of hydration, chemical stability, and even be low in solubility in water. The study is exploring the characteristics of calcium carbonate (CaCO 3 )-modified concrete, such as its physical, rheological, and microstructural properties. CaCO 3 and ground granulated blast furnace slag were taken as a replacement of 5%, 10%, 15% and 20% of the cement by weight, with a constant water to cement ratio being 0.45, and 50% replacement of cement by ground granulated blast furnace slag (GGBS). The plots of Nyquist and Bode that the microstructural analysis of the changes in the concrete matrix. Non-destructive testing was used to test the performance of the concrete in the long term, quality, homogeneity, and the hardness of the surface. The findings revealed that CaCO 3 -modified concrete was better than traditional concrete, especially in its resistance to carbonation. It gave the best result with 15% cement replacement, where compressive strength was 71.1 MPa, split tensile strength was 4.72 MPa, and flexural strength was 5.89 MPa after 90 days. The results indicate that CaCO 3 can be an effective alternative to cement that can be used in the production of concrete.

Rapid ice-marginal lake growth in Alaska driven by glacier retreat through bed overdeepenings

The number and cumulative area of ice-marginal lakes have expanded globally in recent decades, with many lakes residing in glacier-bed overdeepenings, which are subglacial basins formed through preferential glacial erosion. However, current lake expansion rates, key drivers of expansion, and maximum future lake extents are poorly quantified. This is notable because glacial lakes pose hazards, alter hydrologic and ecological systems, and, in some cases, accelerate glacier flow and retreat. Here, we quantify recent ice-marginal lake growth and use existing ice thickness and topographic data to map glacier-bed overdeepenings in Alaska as a predictor of recent and future locations of lake growth. Ice-marginal lakes in the region grew by +156 km 2 (26 km 2 y −1 ) between 2018 and 2024, representing a 50% increase relative to the 2009–2018 rate. Eighty percent of lake growth since 2018 occurred in mapped glacier-bed overdeepenings. Approximately 4,250 km 2 (2,966 to 5,503 km 2 accounting for ± ice thickness uncertainty) of the overdeepened area is connected to an ice-marginal lake, indicating the potential for more than fourfold lake growth of existing lakes as glaciers retreat. An additional 14,500 km 2 (12,469 to 17,134 km 2 ) of remaining glacier area resides on glacier-bed overdeepenings not connected to existing lakes, highlighting the potential for substantial new lake development. Velocities from lake-terminating glaciers show clear passive and dynamic endmembers on a continuum of glacier–lake coupling. Glaciers with ice-marginal lakes thinned 23 to 54% more than glaciers of similar area without lakes, underscoring the critical importance of dynamic glacier–lake coupling on the evolution of glaciers in Alaska.

Abstract MPTU10: Metformin Use, Gene Expression, and Cardiometabolic Health in People with Diabetes: the Hispanic Community Health Study/Study of Latinos (HCHS/SOL)

Background: Metformin may offer anti-aging and cardiometabolic benefits beyond its typical role in glycemic control through regulating gene expression, but its relationship with whole-blood gene expressions remains unclear in human studies. Hypothesis: Metformin use is associated with gene expressions related to anti-aging pathways, and these signatures are associated with favorable cardiometabolic traits. Methods: In 794 participants with diabetes and free of cardiovascular disease and cancer from the HCHS/SOL, differential gene expression analysis was conducted between those with metformin use (n=338) and those without any anti-diabetic medication use (n=456) using whole-blood RNA-sequencing data. Pathway enrichment analysis based on the MSigDB database was conducted for biological interpretation. Spearman correlation analysis was applied to examine correlations between differential expression genes (DEGs) and cardiometabolic traits. Results: Among 14791 genes, we identified 36 upregulated and 43 downregulated DEGs associated with metformin use ( P FDR <0.05; Fig. a ). The gene-set enrichment analysis indicated that metformin use was associated with upregulating heme metabolism and angiogenesis pathways and downregulating apoptosis, coagulation, KRAS signaling, and complement system pathways ( P FDR <0.25; Fig. b ). These alterations indicate enhanced adaptive and repair accompanied by suppressed apoptosis and inflammatory responses, which are important pathways in anti-aging. Some upregulated DEGs, such as YPEL4 (cell cycle progression) and IGHG1 (encodes constant region of IgG1 antibodies) , showed negative correlations with unfavorable traits (e.g., HOMA-IR and CRP), while some downregulated DEGs, such as GNAL (regulating energy metabolism) and C1QB (complement system), showed positive correlations with unfavorable blood lipids ( Fig. d ). However, some DEGs showed unexpected results with cardiometabolic health, with metformin upregulated DEGs (e.g., ARHGEF37 , a gene involved in structure and signal transduction) positively associated with unfavorable cardiometabolic traits. Conclusions: Among Hispanics/Latinos with diabetes, metformin use is associated with alterations in whole-blood gene expressions related to several anti-aging pathways, some of which might be involved in the beneficial effects of metformin use on cardiometabolic health.

Abstract WE468: Axillary/Subclavian IABP Is a Safe and Effective Support in Advanced Heart Failure: A Meta-Analysis

Background: Axillary intra-aortic balloon pumps (IABPs) represent an emerging approach to support patients with advanced heart failure, particularly those awaiting heart transplantation. However, data on their safety and effectiveness remain limited. This study aims to evaluate the potential benefits and adverse effects of axillary IABP through a single-arm meta-analysis. Hypothesis: What are the safety outcomes and clinical benefits of axillary intra-aortic balloon pump (IABP) support in patients with advanced heart failure? Methodology: A systematic search of PubMed, EMBASE, Cochrane Library, and ScienceDirect was conducted to identify observational studies reporting outcomes of axillary or subclavian intra-aortic balloon pump (IABP) placement in patients with advanced heart failure. Primary outcomes were IABP success, mortality, and the need for exchange or repositioning. Secondary outcomes included infection, cerebrovascular accidents (CVA), and limb ischemia. A random-effects model was used to calculate pooled proportions with 95% confidence intervals. Statistical analyses were performed using RStudio with the meta package. Results: We included 9 studies, comprising 584 patients who underwent axillary/subclavian IABP (males: 447; females:138). Pooled mean duration for IABP was 21.49 (95% CI: 14.65-28.33). Etiologically, Ischemic cardiomyopathy was diagnosed in 199 (34%) patients while non ischemic cardiomyopathy in 314 (53.67%) patients. Overall success of IABP was reported in 8 studies which showed the proportion of 0.77 (95% CI: 0.57-0.89). The subgroup analysis showed that, 67% patients proceeded to successful transplant (0.67; 95% CI: 0.58-0.75) while 6% patients were gone through left ventricular assisted device (LVAD). Only 4% death was reported (0.04; 95% CI: 0.01-0.07). The pooled rate of IABP exchange or repositioning was 25% (95% CI: 7%–41%). Re-exploration was done in 6% patients 0.06 (95 % CI: 0.00- 0.11). Pooled proportions for secondary outcomes were: cerebrovascular events 0.02 (95% CI: 0.01–0.04), infection 0.01 (95% CI: 0.00–0.02), and limb ischemia 0.01 (95% CI: 0.00–0.03). Conclusion: Axillary/subclavian IABP is a feasible and effective bridging strategy in advanced heart failure, showing high transplant success and low mortality. Despite a notable rate of device-related complications, serious complications were rare.

Abstract TU213: Human Milk Consumption During the Neonatal Hospitalization is Not Associated with Length- or Weight-for-Age at Discharge for Infants with Critical Congenital Heart Disease.

Introduction: Human milk is the optimal nutritional source for infants. In critically ill infants, human milk offers benefits related to disease reduction (eg, lower necrotizing enterocolitis) and neurodevelopment. Infants with critical congenital heart disease (CCHD), a particularly vulnerable population, may be at risk for below-average perioperative growth. This project aimed to determine whether the percentage of enteral volume received as human milk was associated with length-for-age (LAZ) and weight-for-age (WAZ) z-scores at hospital discharge. We hypothesized a positive association for LAZ and no association for WAZ. Methods: For this retrospective cohort study, daily nutrition information was extracted from the electronic health record (EHR) for infants who underwent neonatal cardiac surgery on cardiopulmonary bypass between 2014–2023. Infants were excluded if their hospital length of stay was <7 days or >180 days, if they received <7 days of enteral feeding, or if they died before discharge. Linear regression was used to estimate the association between the average daily human milk percentage and LAZ or WAZ at hospital discharge. Models were adjusted for sociodemographic and physiological covariates, including infant sex, insurance type, single ventricle physiology, genetic syndrome, and hospital length of stay. Sensitivity analysis was conducted examining human milk consumption as a categorical variable with cutoffs for quartiles 1, 2, 3, and 4 set at 0-25%, 26-50%, 51-75%, and 76%-100%, respectively. Results: Infants (n=211) were 33% female, 71% White, 36% with public insurance, and 42% with single ventricle physiology. Human milk comprised on average 75% of the daily enteral nutrition. Infants’ median (25%,75%) hospital length of stay was 39 (28,56) days, with average -1.53 (±1.49) LAZ and -1.62 (±1.08) WAZ at discharge. There were no significant associations between human milk percentage and LAZ or WAZ at discharge in unadjusted or adjusted models, or in sensitivity analyses (Table 1). Conclusions: Human milk consumption was not associated with differences in discharge LAZ or WAZ among infants with CCHD, suggesting non-inferior growth for infants fed human milk during the neonatal hospitalization for cardiac surgery. Given the well-established benefits of human milk feeding for all infants, these findings support continued human milk promotion for critically ill infants with CCHD.

Euler simulation of hydrocyclone pre-separation before oily wastewater membrane treatment based on Population balance model

Abstract Membrane technology is considered an effective technique for separating and purifying oily wastewater, and membrane fouling caused by surfactant adsorption or oil droplet blockage of pore channels. Therefore, before membrane treatment of oilfield wastewater, it is necessary to carry out pretreatment to reduce the oil content of oilfield wastewater and lay the foundation for improving membrane treatment efficiency. This article mainly uses the finite volume method to discretize the control equations, and uses the pressure velocity coupling method to analyze the primary hydrocyclone separation efficiency of oilfield produced fluids with oil content ranging from 10 to 30%. It explores the hydrocyclone flow field and particle trajectory of mixed fluids in hydrocyclones under different oil content states. The results indicate that the distribution of oil phase volume fraction in the hydrocyclone is comprehensively controlled by inlet flow velocity, oil content, and structural parameters such as cone angle and overflow pipe depth. At low to medium flow velocities (5.0 ~ 15.0 m/s), the centrifugal force field dominates the enrichment of oil towards the axis, and the coalescence effect is enhanced. At high inlet flow rates, the oil phase outflow channel changes from a dual channel of overflow and bottom flow to mainly overflow. When the inlet flow rate reaches 20 m/s, the shear force increases, causing the oil droplets to break and the radial migration path to deviate from the design. The flow velocity distribution becomes more complex, resulting in the distribution characteristics of the oil volume fraction in the mixed liquid becoming more complex, and small-sized oil droplets are discharged from the bottom flow. The increase in oil content enhances the probability of coalescence by increasing viscosity, but also increases Stokes resistance, resulting in prolonged droplet retention time, increased risk of shear failure, and intensified fluctuations in oil phase volume fraction at the overflow port.

Radiocarbon dating and chemical imaging of carbon black–based Paleolithic cave art in the Dordogne region (France)

Paleolithic parietal art in the Dordogne, Southwestern France, was known to present representations solely made with mineral coloring matters. We found a significant number of carbon black-based figures in the galleries of the Font-de-Gaume cave in Les Eyzies, Dordogne, Southern France [I. Reiche, Y. Coquinot, A. Trosseau, A. Maigret, Sci. Rep. 13 , 22235 (2023)]. Further reflectance imaging spectroscopy allowed a precise noninvasive discrimination between manganese- and carbon-based blacks. Consequently, in the Dordogne region, direct dating of drawn or painted lines was unlocked. Dating parietal representations can prove challenging because of the small amount of matter and the possible contaminations by other carbon sources. The sampling was conducted for radiocarbon dating on two selected figures identified as being made with carbon black: the Bison figure HB15 (named by Breuil, today GPCarG-006), located on the left at the level of the Carrefour in the public area of the cave, as well as on the Mask (human or animal face, GL3D-009) on the right of the remote sector 3 of the lateral gallery. Slightly more recent than expected, the obtained dates are of 13461–13162 calBP for the Bison , and of 8993–8590 calBP (left eye), 15981–15121 calBP (upper lip), and 15297–14246 calBP (lower lip) for the Mask . Except for one date, these results represent the experimental confirmation of the Paleolithic age of cave art in the Font-de-Gaume cave. This study opens numerous perspectives for a more systematic dating of the parietal representations of the cavern and motivates further research of carbon black–based Paleolithic parietal art in the Dordogne region.

Abstract MPTU04: Sex- and Disease-Stratified Performance of Federated Learning for Multi-Label Cardiovascular Disease Detection: A Real-World ECG Analysis

Background: Federated learning (FL) enables privacy-preserving collaborative model training across healthcare institutions without sharing patient data. We evaluated disease-specific detection performance and demographic disparities in FL algorithms for ECG-based cardiovascular diagnosis across 20 cardiac conditions. Methods: Using the PTB-XL dataset (n=21,799 ECG recordings, 18,869 patients, 20 cardiac conditions), we trained a ResNet-1D deep learning model for multi-label classification. We assessed performance across critical conditions (atrial fibrillation, myocardial infarctions, AV blocks), rhythm disorders, and conduction abnormalities. Model evaluation included disease-specific AUROC, sensitivity, and clinical priority stratification across sex and age subgroups. Results: On 3,343 independent test samples, the model achieved mean AUROC of 0.853 across 20 conditions. Rhythm disorders showed excellent performance: sinus tachycardia (0.976), right bundle branch block (0.970), atrial flutter (0.970), and sinus bradycardia (0.953). Among myocardial infarctions, anteroseptal MI achieved 0.950 AUROC with 53.9% sensitivity, while inferior MI achieved 0.889 AUROC with 62.8% sensitivity. However, critical limitations emerged: atrial fibrillation showed only 24.7% sensitivity despite 0.881 AUROC, missing 75% of cases—a major clinical concern given stroke risk. Rare conditions suffered from class imbalance: third-degree AV block (n=4, 0% sensitivity) and second-degree AV block (n=2, 0% sensitivity). Demographic analysis revealed minimal sex disparity (0.09% accuracy difference) but age-related performance decline in elderly populations (>80 years). Conclusion: Deep learning achieves strong disease-specific detection for common cardiovascular conditions (AUROC >0.95 for rhythm disorders), but class imbalance severely limits rare condition detection. Low atrial fibrillation sensitivity represents a critical safety gap requiring intervention strategies. Federated learning offers promise for multi-site collaborative diagnosis while preserving patient privacy, though disease-specific validation is essential. Clinical Implications: Models excel at detecting frequent conditions but fail on rare, life-threatening diseases. Clinical deployment requires condition-specific performance thresholds, class balancing strategies (oversampling, focal loss), and mandatory rare condition review protocols.

Abstract 65: Food Insecurity Modifies Effect of GoFresh DASH-Patterned Grocery Intervention on Blood Pressure in Black Adults with High Blood Pressure

Introduction: While the DASH diet improves blood pressure (BP) in tightly controlled research studies, health-related social needs such as food insecurity may limit its implementation in real-world settings. How food insecurity status affects BP impacts from DASH-type diets is unknown. Hypothesis: A DASH-patterned grocery delivery intervention will more effectively lower BP for adults experiencing food insecurity. Methods: The GoFresh study randomized 180 Black adults with high BP to receive either weekly dietitian-supported DASH-patterned grocery deliveries or $500 stipends every 4 weeks for 12 weeks. Food security status was assessed at baseline using the 2-item Hunger Vital Sign tool. Systolic BP (SBP) and diastolic BP (DBP) were measured at baseline and 3-months. We used generalized estimating equations to assess differences in BP changes by baseline food security status and study arm. Results: Of 180 participants, 102 were experiencing food insecurity and 78 were not ( Table 1 ). Overall, participants with food insecurity showed greater BP reductions with the intervention compared to participants without (p-interaction=0.15 for SBP; p-interaction=0.03 for DBP) ( Figure 1 ). Among participants with food insecurity, SBP changed by -6.66 mmHg (95%CI -9.00, -4.31) in the intervention group vs -1.59 mmHg (95%CI -3.97, 0.78) in the self-directed group. Similarly, DBP changed by -4.93 mmHg (95%CI -6.61, -3.26) in the intervention group vs -0.80 mmHg (95%CI -2.49, 0.89) in the self-directed group. Among participants who were food secure, SBP changed by -4.44 mmHg (95%CI -7.19, -1.69) in the intervention group and -3.11 mmHg (95%CI -5.76, -0.45) in the self-directed group. DBP changed by -2.58 mmHg (95%CI -4.55, -0.62) in the intervention group and -2.43 mmHg (95%CI -4.32, -0.54) in the self-directed group. Participants with food insecurity saw greater SBP reduction with the intervention than those who were food secure (-3.73 mmHg; 95%CI: -8.80, 1.35), although not statistically significant. Similarly, participants with food insecurity showed significantly greater DBP reduction with the intervention (-3.98 mmHg; 95%CI -7.60, -0.35) compared to those who were food secure ( Table 2 ). Conclusions: DASH-patterned grocery delivery reduced DBP more among participants experiencing food insecurity than those who were not. These findings highlight the importance of addressing food insecurity to enhance cardioprotective effects of dietary interventions.

Abstract WE552: Differences in metabolic, behavioural and psychosocial risk factors and mortality in women and men: nationwide population based cohort study

Background: Little is known about the interaction and joint associations of sex and modifiable risk factors with mortality, leaving critical gaps in understanding their contribution to population health outcomes. Methods: The Prospective Urban Rural Epidemiology (PURE)-China cohort study recruited participants aged 35 to 70 years from 115 urban and rural communities across 12 provinces in China during 2005-2009, and followed up until Aug 30, 2021. Metabolic, behavioural, and psychosocial risk factors were assessed using standardized measures and questionnaires, with all-cause mortality serving as the primary outcome. Results: Of 47,345 participants included in this analysis, 27,584 (58.3%) were women. After a median follow-up of 11.9 years (IQR 9.5-12.6), 1217 deaths were observed in women (age-standardised mortality 3.8 [95% CI 3.6-4.0] per 1000 person-years) and 1526 in men (6.1 [5.8-6.4] per 1000 person-years). Men exhibited higher prevalence of metabolic, behavioural, and psychosocial risk factors, although these differences diminished with age. Systolic blood pressure showed a stronger association with mortality in men (HR 1.25 [95% CI 1.19-1.31] per 20 mm Hg) than that among women (1.14 [1.08-1.20] per 20 mm Hg, P interaction <0.001), as was the case for hypertension. Conversely, women with low physical activity (<600 metabolic equivalents×minutes per week or <150 minutes per week of moderate intensity physical activity) (1.29 [1.10-1.51] vs. 1.00 [0.87-1.15], P interaction =0.003) and a PURE diet score ≤3 (1.18 [1.02-1.37] vs. 1.02 [0.90-1.15], P interaction =0.002) were associated with higher risks of mortality than that among men. Men with ≥7 risk factors had a significantly greater risk of mortality compared to women with 0-2 risk factors (3.40 [2.71-4.27]). Metabolic risk factors accounted for a higher excessive mortality in men (PAF 26.2% vs. 15.4% in women), whereas behavioural and psychosocial factors contributed more in women (35.1% vs. 24.5% in men). Conclusions: These findings underscore that metabolic risk factors, particularly systolic blood pressure, are key drivers of mortality in men, while behavioural and psychosocial factors, including physical inactivity and poor diet, are more influential in women. Targeted interventions addressing these sex-specific risk factors could significantly reduce mortality and enhance health equity in the Chinese population.

Computational investigation on the hydrodynamic performance of a vertically submerged plate-type wave energy converter under variable relative openings