medRxiv : serveur de preprint pour les sciences médicales

Eating disorders (EDs) are a type of psychiatric disorder characterized by pathological eating and related behavior and considered to be highly heritable. The purpose of this study was to explore rare variants expected to display biological functions associated with the etiology of EDs. We performed whole exome sequencing (WES) of affected sib-pairs corresponding to disease subtype through their lifetime and their parents. From those results, rare single nucleotide variants (SNVs) concordant with sib-pairs were extracted and estimated to be most deleterious in the examined families. Two non-synonymous SNVs located on corticotropin releasing hormone receptor 2 (CRHR2) and glutamate metabotropic receptor 8 (GRM8) were identified as candidate disease susceptibility factors. The SNV of CRHR2 was included within the cholesterol binding motif of the transmembrane helices region, while the SNV of GRM8 was found to contribute to hydrogen bonds for an -helix structure. CRHR2 plays important roles in the serotoninergic system of dorsal raphe nuclei, which is involved with feeding and stress-coping behavior. Moreover, GRM8 modulates glutamatergic neurotransmission, and is also considered to have effects on dopaminergic and adrenergic neurotransmission. Further investigation regarding the biological function of these variants may provide an opportunity for elucidate the pathogenesis of EDs.

Multiple attempts at intracranial hemorrhage (ICH) detection using deep-learning techniques have been made and plagued with clinical failures. Most studies for ICH detection have insufficient data or weak annotations. We sought to determine whether a deep-learning algorithm for ICH detection trained on a strongly annotated dataset outperforms that trained on a weakly annotated dataset, and whether a weighted ensemble model that integrates separate models trained using datasets with different ICH subtypes is more accurate. We used publicly available brain CT scans from the Radiological Society of North America (27,861 CT scans, 3,528 ICHs) and AI-Hub (53,045 CT scans, 7,013 ICHs) for training datasets. For external testing, 600 CT scans (327 with ICH) from Dongguk University Medical Center and 386 CT scans (160 with ICH) from Qure.ai were used. DenseNet121, InceptionResNetV2, MobileNetV2, and VGG19 were trained on strongly and weakly annotated datasets and compared. We then developed a weighted ensemble model combining separate models trained on all ICH, subdural hemorrhage (SDH), subarachnoid hemorrhage (SAH), and small-lesion ICH cases. The final weighted ensemble model was compared to four well-known deep-learning models. Six neurologists reviewed difficult ICH cases after external testing. InceptionResNetV2, MobileNetV2, and VGG19 models outperformed when trained on strongly annotated datasets. A weighted ensemble model combining models trained on SDH, SAH, and small-lesion ICH had a higher AUC than a model only trained on all ICH cases. This model outperformed four well-known deep-learning models in terms of sensitivity, specificity, and AUC. Strongly annotated data are superior to weakly annotated data for training deep-learning algorithms. Since no model can capture all aspects of a complex task well, we developed a weighted ensemble model for ICH detection after training with large-scale strongly annotated CT scans. We also showed that a better understanding and management of cases challenging for AI and human is required to facilitate clinical use of ICH detection algorithms.

Background: We investigated systemic biochemical changes in Alzheimer's disease (AD) by investigating the relationship between circulating plasma metabolites and both clinical and biomarker-assisted diagnosis of AD. Methods: We used an untargeted approach with liquid chromatography coupled to high-resolution mass spectrometry to measure exogenous and endogenous small molecule metabolites in plasma from 150 individuals clinically diagnosed with AD and 567 age-matched elderly without dementia of Caribbean Hispanic ancestry. Plasma biomarkers of AD were also measured including P-tau181, A{beta}40, A{beta}42, total tau, neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP). Association of individual and co-expressed modules of metabolites were tested with the clinical diagnosis of AD, as well as biologically-defined AD pathological process based on P-tau181 and other biomarker levels. Results: Over 4000 metabolomic features were measured with high accuracy. First principal component (PC) of lysophosphatidylcholines (lysoPC) that bind to or interact with docosahexaenoic acid (DHA), eicosapentaenoic acid (EPA) and arachidonic acid (AHA) was associated with decreased risk of AD (OR=0.91 [0.89-0.96], p=2e-04). Restricted to individuals without an APOE {varepsilon}4 allele (OR=0.89 [0.84-0.94], p= 8.7e-05), the association remained. Among individuals carrying at least one APOE {varepsilon}4 allele, PC4 of lysoPCs moderately increased risk of AD (OR=1.37 [1.16-1.6], p=1e-04). Essential amino acids including tyrosine metabolism pathways were enriched among metabolites associated with P-tau181 levels and heparan and keratan sulfate degradation pathways were associated with A{beta}42/A{beta}40 ratio reflecting different pathways enriched in early and middle stages of disease. Conclusions: Our findings indicate that unbiased metabolic profiling can identify critical metabolites and pathways associated with {beta}-amyloid and phosphotau pathology. We also observed an APOE {varepsilon}4 dependent association of lysoPCs with AD and that biologically-based diagnostic criteria may aid in the identification of unique pathogenic mechanisms.

The decision to accept an organ offer for transplant, or wait for something potentially better in the future, can be challenging. Especially, clinical decision support tools predicting transplant outcomes are lacking. This project uses interpretable methods to predict both graft failure and patient death using data from previously accepted kidney transplant offers. Precisely, using more than twenty years of transplant outcome data, we train and compare several survival analysis and classification models in both single and multiple risk settings. In addition, we use post hoc interpretability techniques to clinically validate these models. In a single risk setting, neural networks provide comparable results to the Cox proportional hazard model, with 0.71 and 0.81 AUROC for predicting graft failure and patient death at year 10, respectively. Recipient and donor ages, primary renal disease, donor eGFR, donor type, and the number of mismatches at DR locus appear to be important features for transplant outcome prediction. We also extended the neural network approach to multiple outcome prediction, maintaining consistent performances and clinical interpretation. Thus, owing to their good predictive performance and the clinical relevance of their post hoc interpretation, neural networks represent a promising core component in the construction of future decision support systems for transplant offering.

Introduction Between March 2020-June 2021, over 30 million COVID-19 cases were reported in India. We assessed the COVID-19 response across the US Mission India (US Embassy New Delhi, US Consulates - Mumbai, Hyderabad, Chennai and Kolkata) to plan future mitigation efforts and fill gaps in knowledge about COVID-19 transmission in a unique community like the US Mission. Method We described COVID-19 mitigation activities undertaken by the five US Mission India posts and conducted a secondary analysis of case investigation and contact tracing program data collected by the Health Unit from March 2020-July 2021. Results US Mission in India, in collaboration with multiple internal agencies, initiated COVID-19 mitigation activities in March 2020. Activities included educational sessions, training for infection prevention and control, health and safety assessments, and the development of standard operating procedures (SOPs). The Health Unit and US CDC India office initiated COVID-19 case investigations and conducted contact tracing. Between March 2020-July 2021, 636 COVID-19 cases (72% males), including 48 clusters (size range 2-10 cases), were reported. Overall case fatality rate was 1.5%. Of case patients, 82% (523) were Indians, and 18% (113) were Americans. On presentation, 22% (138/625) of cases were asymptomatic. The median time from symptom onset to notification to the Health Unit was three days (Interquartile range 1-5). The Health Unit identified 2,484 contacts (positivity rate 25%). Frequency of case presentation in the US Mission India closely resembled the pattern of COVID-19 waves in India. The attack rates ranged over the time period between 10-19%, the highest at 19% in Delhi. Conclusions COVID-19 mitigation strategies were implemented in collaboration with multiple agencies and helped prevent the transmission of COVID-19 and large COVID-19 clusters in the US Mission India.

The uptake of COVID-19 vaccines remains low despite their high effectiveness. Epidemic models that represent decision-making psychology can provide insight into the potential impact of vaccine promotion interventions in the context of the COVID-19 pandemic. We coupled a network-based mathematical model of SARS-CoV-2 transmission in Georgia, USA with a social-psychological vaccination decision-making model in which vaccine side effects, post-vaccination infections, and other unidentified community-level factors could "nudge" individuals towards vaccine resistance while hospitalization spikes could nudge them towards willingness. Combining an increased probability of hospitalization-prompted resistant-to-willing switches with a decreased probability of willing-to-resistant switches prompted by unidentified community-level factors increased vaccine uptake and decreased SARS-CoV-2 incidence by as much as 30.7% and 24.0%, respectively. The latter probability had a greater impact than the former. This illustrates the disease prevention potential of vaccine promotion interventions that address community-level factors influencing decision-making and anticipate the case curve instead of reacting to it.

Cardiovascular disease (CVD) prevalence in Kenya is rising. Overweight, pre-hypertension and physical inactivity in younger ages is contributory. These risk factors are inadequately documented among Kenyan children and adolescents, hampering CVD prevention. This cross-sectional study randomly sampled 384 school children and adolescents. After obtaining assent and consent, global physical activity (PA) questionnaire was used to assess PA. Body mass index (BMI), Waist-Hip Ratio (WHR) and Waist-Height Ratio (WHtR) were determined. Blood pressure (BP) was also measured. Participants were 14.6{+/-}2.7 years, and 62.6% were female. Overall BMI was 19.8{+/-}3.9 kg/m2 with 8% having [≥]25.0 kg/m2, 87% of whom were in secondary schools. Using SBP, 27.9% were at risk of CVD. For age [≥]13 years old, males and females proportions were 42.5% and 20% respectively while that for <13 years old, it was 26.5% and 27% respectively. For DBP, 12.8% had elevated-to-hypertensive BP. For ages [≥] 13 years old, this was 13.2% and 8.3% for males and females respectively and for the <13 years old, the respective percentages were 11.8 and 25.4. Combining SBP and DBP, 8.1% of participants, mostly males, had elevated-to-hypertensive BP. Thirty-one percent of boys and 15.6% of girls were at CVD risk using respective WHR cutoffs of 0.90 and 0.85 (males, 0.93{+/-}0.02 ; females, 0.89{+/-}0.03). For WHtR, 39.6% of boys had values >0.463 cut-off (0.493{+/-}0.02), with 52.6% in secondary schools against 32.4% for girls having >0.469 cut-off (0.517{+/-}0.05), 69.7% being in secondary schools. Overall, 45% of participants were sports-inactive and 77.2% did minimal PA. Among school-going children and adolescents in Eldoret, Kenya, prevalence of CVD risk-factors was high especially among boys and in high schools. Large proportions had elevated BP, BMI, WHR and WHtR, and, further, were sedentary, posing high CVD risk. Lifestyle interventions to mitigate this public health concern among children and adolescent are urgently needed.

Xylazine is increasingly reported in street drugs and fatal overdoses in the United States (US), often in combination with synthetic opioids, yet state-level xylazine data are limited, hampering local public health responses. The present study analyzed 2018-2022 state-level data from the National Forensic Laboratory Information System (xylazine-positive reports of seized drugs analyzed by forensic laboratories), the Centers for Disease Control and Prevention (population estimates, synthetic opioid overdose mortality rates), and individual states' medical examiner/public health agency reports (numbers of xylazine-involved overdose deaths). An ordinary least squares regression model predicted state-level synthetic opioid overdose mortality rates by xylazine seizure report rates, adjusting for US Census Region. In 2018, relatively low rates of xylazine seizure reports were observed, with 21 states reporting zero xylazine seizures. In 2022, only three states reported zero xylazine seizures, and the highest xylazine seizure report rates (per 100,000 residents) were observed in New Jersey (30.4), Rhode Island (22.7), Maryland (18.9), Virginia (15.5), New Hampshire (13.0), and Ohio (10.9). Data on 2019-2022 xylazine-involved overdose deaths were available for 21 states/DC (60 state-years), with the highest 2022 xylazine-involved overdose death rates (per 100,000 residents) in Vermont (10.5) and Connecticut (9.8). Finally, in 2021, at the state level, each additional reported xylazine seizure per 100,000 residents was associated with a 2% higher synthetic opioid overdose mortality rate (b=0.02, robust standard error=0.01; p=0.049). Overall, study results emphasize xylazine's increasing involvement in US law enforcement drug seizure reports and overdose deaths, primarily in the East, yet also extending across the country.

Waning antibody responses after COVID-19 vaccination combined with the emergence of the SARS-CoV-2 Omicron lineage led to reduced vaccine effectiveness. As a countermeasure, bivalent mRNA-based booster vaccines encoding the ancestral spike protein in combination with that of Omicron BA.1 or BA.5 were introduced. Since then, BA.2-descendent lineages have become dominant, such as XBB.1.5 or BA.2.86. Here, we assessed how different COVID-19 priming regimens affect the immunogenicity of the recently used bivalent booster vaccinations and breakthrough infections. BA.1 and BA.5 bivalent vaccines boosted neutralizing antibodies and T-cells up to 3 months after boost; however, cross-neutralization of XBB.1.5 was poor. Interestingly, different combinations of prime-boost regimens induced divergent responses: participants primed with Ad26.COV2.S developed lower binding antibody levels after bivalent boost while neutralization and T-cell responses were similar to mRNA-based primed participants. In contrast, the breadth of neutralization was higher in mRNA-primed and bivalent BA.5 boosted participants. Combined, we highlight important "lessons learned" from the employed COVID-19 vaccination strategies. Our data further support the use of monovalent vaccines based on circulating strains when vaccinating risk groups, as recently recommended by the WHO. We emphasize the importance of the continuous assessment of immune responses targeting circulating variants to guide future COVID-19 vaccination policies.

Importance: Streptococcus pneumoniae is a known etiology of acute respiratory infections (ARIs), which account for large proportions of outpatient visits and antibiotic use in children. In 2023, 15- and 20-valent pneumococcal conjugate vaccines (PCV15, PCV20) were recommended for routine use in infants. However, the burden of outpatient healthcare utilization among U.S. children attributable to the additional, non-PCV13 serotypes in PCV15/20 is unknown. Objective: To estimate the incidence of outpatient visits and antibiotic prescriptions in U.S. children for acute otitis media, pneumonia, and sinusitis associated with PCV15- and PCV20-additional serotypes (non-PCV13 serotypes) to quantify potential impacts of PCV15/20 on outpatient visits and antibiotic prescriptions for these conditions. Design: Multi-component study including descriptive analyses of cross-sectional and cohort data on outpatient visits and antibiotic prescriptions from 2016-2019 and meta-analyses of pneumococcal serotype distribution in non-invasive respiratory infections. Setting: Outpatient visits and antibiotic prescriptions among U.S. children. Participants: Pediatric visits and antibiotic prescriptions among children captured in the National Ambulatory Medical Care Survey (NAMCS), the National Hospital Ambulatory Medicare Care Survey (NHAMCS), and Merative MarketScan, collectively representing healthcare delivery across all outpatient settings. Incidence denominators estimated using census (NAMCS/NHAMCS) and enrollment (MarketScan) data. Main outcome(s) and measure(s): Pediatric outpatient visit and antibiotic prescription incidence for acute otitis media, pneumonia, and sinusitis associated with PCV15/20-additional serotypes. Results: We estimated that per 1000 children annually, PCV15-additional serotypes accounted for 2.7 (95% confidence interval 1.8-3.9) visits and 2.4 (1.6-3.4) antibiotic prescriptions. PCV20-additional serotypes resulted in 15.0 (11.2-20.4) visits and 13.2 (9.9-18.0) antibiotic prescriptions annually per 1,000 children. Projected to national counts, PCV15/20-additional serotypes account for 173,000 (118,000-252,000) and 968,000 (722,000-1,318,000) antibiotic prescriptions among U.S. children each year, translating to 0.4% (0.2-0.6%) and 2.1% (1.5-3.0%) of all outpatient antibiotic use among children. Conclusions and relevance: PCV15/20-additional serotypes account for a large burden of pediatric outpatient healthcare utilization. Compared with PCV15-additional serotypes, PCV20-additional serotypes account for >5 times the burden of visits and antibiotic prescriptions. These higher-valency PCVs, especially PCV20, may contribute to preventing ARIs and antibiotic use in children.

Abstract: Background: Astrovirus is a leading cause of acute gastroenteritis in children worldwide. However, few prospective studies have analyzed astrovirus in community-dwelling pediatric populations in low-and-middle-income countries. Methods: We assessed the incidence, risk factors, clinical characteristics, genotypes, viral coinfections and seasonality of astrovirus gastroenteritis in 443 healthy Nicaraguan children born in 2017-2018, followed for 36 months. Children were recruited from maternity hospitals and birth records in an economically-diverse neighborhood of Leon, the second-largest city in Nicaragua. Astrovirus-positive episodes and genotypes were identified from diarrheal specimens with reverse transcription quantitative polymerase chain reaction and Sanger sequencing. Results: Of 1708 total specimens tested, eighty children (18%) experienced at least 1 astrovirus episode, and 9 experienced repeat episodes, mostly during the rainy season (May-October). The incidence of astrovirus episodes was 7.8/100 child-years (95% CI: 6.2, 9.8). Genotype-specific incidence of astrovirus also exhibited seasonality. Median age of astrovirus episode onset was 16 months (IQR 9, 23). Initial astrovirus episodes were not associated with protection against future episodes during the age span studied. Astrovirus cases were exclusively breastfed for a shorter period than uninfected children, and the human milk oligosaccharide lacto-N-fucopentaose-I was more concentrated in mothers of these children. Home toilets appeared to protect against future astrovirus episodes (HR=0.19, 95% CI 0.04-0.91). Human astrovirus-5 episodes, comprising 15% of all typed episodes, were associated with longer diarrhea and more symptomatic rotavirus co-infections. Conclusion: Astrovirus was a common cause of gastroenteritis in this cohort, and future studies should clarify the role of astrovirus genotype in clinical infection severity.

Togo is a low-income country in West Africa. Estimates suggest that only 25% of the Togolese population have received at least one dose of any COVID-19 vaccine by June 2023. Whilst the early phase of the pandemic vaccine rollout across 2021 was dominated by higher-income countries taking much of the available supply, there have long been sufficient supplies for all nations. Thus, there remains a need to understand reasons for low uptake in countries such as Togo. Two cross-sectional telephone surveys of Togo residents were conducted in December 2020 and January 2022. These surveys asked questions around perceptions of COVID-19, trust in public health messaging, belief in conspiracy theories, and hesitancy around COVID-19 vaccination. Analyses here focus on unvaccinated respondents. Across Survey 1 (N = 1430) and Survey 2 (N = 212), 65% of respondents were men, 47% lived in Lome (capital city of Togo), 25% completed higher education, 67% were married, and 69% were Christian. Between Surveys 1 and 2, overall hesitancy (33.0% to 58.0%) and beliefs in conspiracy theories (29% to 65%) significantly increased. Using logistics regression, governmental mistrust was the strongest significant predictor of hesitancy (OR: 2.90). Participants who indicated agreement or uncertainty with at least one conspiracy belief also predicted greater vaccine hesitancy (OR: 1.36). Proactive approaches to public health messaging, that better understand reasons for hesitancy across different demographics, can support uptake of COVID-19 vaccinations within Togo. This includes health promotion campaigns that use locally and nationally trusted knowledge providers (e.g. the health service or religious leaders) for greatest effectiveness at reducing impact of misinformation. Key future research should focus around knowledge gaps and areas of mistrust created by the pandemic, such as the impact of misinformation upon routine immunisation uptake.

Using data on 119 low- and lower-middle income countries from 2002 to 2020, we apply fixed-effects (FE) methods to evaluate the impacts of three different classifications of development assistance on access to three types of sexual and reproductive health (SRH) services: access to skilled birth attendance, prevalence of modern contraceptives, and coverage of antiretroviral therapies against HIV/AIDS. The results suggest that aid has had a small, but positive effect on these outcomes over this period. For example, SRH aid has increased service coverage rates by between 0.190 and 0.628 percentage points. The results also indicate that the effect of aid has improved across the period and is larger in low-income countries compared with lower-middle income countries. The findings also indicate that bilateral aid may be somewhat more effective than other types of aid. Importantly, the results suggest that development assistance is more effective if it reaches a certain share of overall health spending. The findings are robust to a series of sensitivity checks. The results of the study lend support to the continued allocation of aid to low-income countries to support the provision of sexual and reproductive health services. Both providers and recipients of SRH aid would be advised to identify ways to improve the effectiveness of development assistance in this area.

Objective: To propose a novel approach for enhancing clinical prediction models by combining structured and unstructured data with multimodal data fusion. Methods: We presented a comprehensive framework that integrated multimodal data sources, including textual clinical notes, structured electronic health records (EHRs), and relevant clinical data from National Electronic Injury Surveillance System (NEISS) datasets. We proposed a novel hybrid fusion method, which incorporated state-of-the-art pre-trained language model, to integrate unstructured clinical text with structured EHR data and other multimodal sources, thereby capturing a more comprehensive representation of patient information. Results: The experimental results demonstrated that the hybrid fusion approach significantly improved the performance of clinical prediction models compared to traditional fusion frameworks and unimodal models that rely solely on structured data or text information alone. The proposed hybrid fusion system with RoBERTa language encoder achieved the best prediction of the Top 1 injury with an accuracy of 75.00% and Top 3 injuries with an accuracy of 93.54%. Conclusion: Our study highlights the potential of integrating natural language processing (NLP) techniques with multimodal data fusion for enhancing clinical prediction models' performances. By leveraging the rich information present in clinical text and combining it with structured EHR data, the proposed approach can improve the accuracy and robustness of predictive models. The approach has the potential to advance clinical decision support systems, enable personalized medicine, and facilitate evidence-based health care practices. Future research can further explore the application of this hybrid fusion approach in real-world clinical settings and investigate its impact on improving patient outcomes.

Background: Physical performance tests are predictive of mortality and have been proposed for screening for certain health conditions (e.g., sarcopenia); however, the diagnostic screening and prognostic value of physical performance tests has primarily been studied in age-limited or disease-specific cohorts. In this study, we sought to identify the most salient characteristics associated with three lower quarter balance and strength tests in a deeply phenotyped cohort of community-dwelling adults. Methods: We applied a stacked elastic net approach on detailed data on sociodemographic, health and health-related behaviors, and biomarker data from the first visit of the Project Baseline Health Study (N=2502) to determine which variables were most associated with three physical performance measures: single-legged balance test (SLBT), sitting-rising test (SRT), and 30-second chair-stand test (30CST). Analyses were stratified by age (<65 and [≥]65). Results: Female sex, Black or African American race, lower educational attainment, and health conditions such as non-alcoholic fatty liver disease and cardiovascular conditions (e.g., hypertension) were consistently associated with worse performance across all three tests. Several other health conditions were associated with either better or worse test performance, depending on age group and test. C-reactive protein was the only laboratory value associated with performance across age and test groups with some consistency. Conclusions: Our results highlighted previously identified and several novel salient factors associated with performance on the SLBT, SRT, and 30CST. Future research should discern and validate the value of these tests as affordable, noninvasive biomarkers of prevalent and/or future disease in the community.

Background: The causal relationship between maternal smoking in pregnancy and reduced offspring birth weight is well established and is likely due to impaired placental function. However, observational studies have given conflicting results on the association between smoking and placental weight. We aimed to estimate the causal effect of newly pregnant mothers quitting smoking on their placental weight at the time of delivery. Methods: We used one-sample Mendelian randomization, drawing data from the Avon Longitudinal Study of Parents and Children (ALSPAC) (up to N = 805) and the Norwegian Mother, Father and Child Cohort Study (MoBa) (up to N = 4475). The analysis was performed in pre-pregnancy smokers only, due to the specific role of the genetic instrument SNP rs1051730 (CHRNA5-CHRNA3-CHRNB4) in affecting smoking cessation but not initiation. Results: Fixed effect meta-analysis showed a 175 g [95%CI: 16, 334] higher placental weight for pre-pregnancy smoking mothers who continued smoking at the beginning of pregnancy, compared with those who stopped smoking. Using the number of cigarettes smoked per day in the first trimester as the exposure, the causal estimate was a 12 g [95%CI: 2,22] higher placental weight per cigarette per day. Results were similar when the smoking exposures were measured at the end of pregnancy. Using the residuals of birth weight regressed on placental weight as the outcome, we showed weak evidence of lower offspring birth weight relative to the placental weight for continuing smoking. Conclusion: Our results suggest that continued smoking during pregnancy causes higher placental weights.

Background Extreme but discrete fine particle <2.5m (PM2.5) exposure is associated with higher prevalence of respiratory symptoms. It is unknown whether these effects abate, persist, or worsen over time, nor whether COVID-19 exacerbates PM2.5 effects. Methods We analysed longitudinal survey data from a cohort residing near a 2014 coalmine fire in regional Australia. A 2016/2017 survey included n=4,056 participants, of whom n=612 were followed-up in 2022. Items include questions about 7 respiratory symptoms, history of COVID-19, and time-location diaries that were combined with geospatial models of fire-related PM2.5. Associations were examined using logistic and mixed-effects logistic regressions. Results PM2.5 exposure predicted higher prevalence of chronic cough and current wheeze 2-3 years post-fire. At the 2022 follow-up, PM2.5 exposure was associated with worsening prevalence of chronic cough and possibly current wheeze. While were no detectable interaction effects between PM2.5 and COVID-19, participants with a history of COVID-19 exhibited more significant associations between PM2.5 exposure and respiratory symptoms. Discussion Short-term but extreme PM2.5 may increase the long-term prevalence of chronic cough, while COVID-19 may exacerbate the effect on other respiratory symptoms.

BACKGROUND Dysphagia after stroke is associated not only with poor outcome and higher mortality, but also with reduced quality of life and social isolation. We aimed to investigate the potential relationship between swallowing impairment and symptoms of anxiety and depression after ischemic stroke. METHODS Consecutive patients with ischemic stroke participating in the prospective STROKE-CARD Registry study at the study center Innsbruck, Austria from 2020 to 2022 were assessed for presence of dysphagia on hospital admission (clinical swallowing assessment) as well as for persistence until hospital discharge and a 3-month follow-up (SINGER Independency Index). Symptoms of anxiety and depression were recorded using Beck Depression Inventory (BDI) and Hospital Anxiety and Depression Scale (HADS) at 3-month follow-up. RESULTS Of 648 patients (36.6% female), 19.3% had dysphagia at hospital admission which persisted in 14.8% (hospital discharge) and 6.8% (3-month follow-up). With the presence or duration of dysphagia (no dysphagia, dysphagia at baseline, discharge or at 3 months), score points (mean{+/-}SD) increased in the BDI (7.9{+/-}6.7, 12.5{+/-}8.7, 13.5{+/-}9.0, 16.5{+/-}10.2), HADS-D (4.4{+/-}3.7, 7.1{+/-}4.2, 7.7{+/-}4.4, 9.8{+/-}4.3) and HADS-A (4.4{+/-}3.5, 5.4{+/-}3.6, 6.0{+/-}3.6, 7.0{+/-}3.6). In linear regression analysis adjusting for age, sex and functional disability, BDI and HADS-D but not HADS-A scores were significantly higher in patients with dysphagia when compared to those without dysphagia at baseline or who recovered to discharge and follow-up. Moreover, patients with swallowing impairment were more likely to receive antidepressants, antipsychotics or benzodiazepines at discharge and 3-month follow-up. CONCLUSIONS Dysphagia after stroke is common and severely affects psychosocial functioning of individuals. Our results highlight swallowing impairment as an independent predictor of depressive symptoms after stroke.

The growing aging population means that increasing numbers of people are living with Parkinsons disease (PD). Although dopamine replacement is effective, there are currently no curative or disease-modifying treatments. Nutrition is a well-known environmental risk factor of PD. This study aimed to clarify the relationship between nutritional traits and the prodromal phase of PD. Subjects were enrolled from community-dwelling older adults. To identify individuals with prodromal PD (PPD), probability of PPD (PPP) and mild parkinsonian sign (MPS) assessments were used. Nutritional status was evaluated using a self-administered food frequency questionnaire form. Intakes of total energy, the three major nutrients (fat, protein, and carbohydrate), B vitamins, water, and dietary fiber were calculated. Nutrient intake in PPD was analyzed using analysis of covariance with age and sex as covariates. We enrolled 305 subjects. The prevalence of older adults living with PPD was 4.1% by PPP [≥] 0.3 and 21.0% by MPS criteria. Compared with subjects without PPD, individuals with PPP [≥] 0.3 had significantly higher fat intake (33.4 {+/-} 8.8 vs. 29.3 {+/-} 6.7 g/1000 kcal/day, p=0.037) and lower carbohydrate intake (121.4 {+/-} 19.0 vs. 132.1 {+/-} 17.7 g/1000 kcal/day, p=0.039), and MPS-positive subjects had significantly lower total energy intake (1817.5 {+/-} 384.7 vs. 1933.7 {+/-} 433.7 kcal/day, p=0.012). In this community-based cross-sectional study, nutritional traits differed between older individuals with and without PPD; these differences were more pronounced when the PPP was used to evaluate PPD. Nutrition and diet may thus modify PD incidence.

Early detection of Alzheimer's Disease (AD) is crucial to ensure timely interventions and optimize treatment outcomes for patients. While integrating multi-modal neuroimages, such as MRI and PET, has shown great promise, limited research has been done to effectively handle incomplete multi-modal image datasets in the integration. To this end, we propose a deep learning-based framework that employs Mutual Knowledge Distillation (MKD) to jointly model different sub-cohorts based on their respective available image modalities. In MKD, the model with more modalities (e.g., MRI and PET) is considered a teacher while the model with fewer modalities (e.g., only MRI) is considered a student. Our proposed MKD framework includes three key components: First, we design a teacher model that is student-oriented, namely the Student-oriented Multi-modal Teacher (SMT), through multi-modal information disentanglement. Second, we train the student model by not only minimizing its classification errors but also learning from the SMT teacher. Third, we update the teacher model by transfer learning from the student's feature extractor because the student model is trained with more samples. Evaluations on Alzheimer's Disease Neuroimaging Initiative (ADNI) datasets highlight the effectiveness of our method. Our work demonstrates the potential of using AI for addressing the challenges of incomplete multi-modal neuroimage datasets, opening new avenues for advancing early AD detection and treatment strategies.