Hikikomori, or prolonged social withdrawal, is an issue of global relevance. The HRI-15 is a brief tool for its assessment. This study enhances its utility by adding clinical thresholds to identify risk and person-centered clinical profiles, and by developing a machine-learning(ML)-based computerized adaptive test (CAT) to enable rapid screening. Data from a national survey of Italian adolescents (N = 8,755) were used to conduct ROC analysis and latent profile analysis (LPA). The findings indicated that a score of ≥ 42 achieved optimal classification performance, and four profiles with distinct meanings and systematic differences in anxiety, depression, impulsivity, and risk behaviors were identified. A multivariate conditional inference tree was estimated to develop an ML-based adaptive version of the instrument. The CAT reduced administered items by 53% (7.03/15), accurately reproducing full-length scores (r = .77-.997) and the corresponding classification; alignment was assessed with latent transition analysis (entropy = .96). The procedure was integrated into an application supporting administration, scoring, and reporting (score, risk, profiles, graphs). Combining a cross-validated cutoff and clinically useful profiles with CAT and its automated administration and reporting application strengthens triage and personalization, enabling large-scale, multidomain screening and earlier intervention for hikikomori risk.

Childbirth occurring at high-risk maternal ages (≤18 years or ≥40 years) remains an important public health concern in Bangladesh and is influenced by a range of socio-demographic and behavioral factors. We conducted a cross-sectional secondary analysis of 15,386 ever married women from the BDHS 2022 dataset. After cleaning and removing duplicates and missing values, data were split into training 80 percent and testing 20 percent sets. Three feature selection methods Lasso with 12 features, Chi Square with 13 features, and Boruta with 10 features were applied. Machine learning models including non-ensemble models CART, SVM, KNN, Naive Bayes and ensemble models Random Forest, GBM, AdaBoost, XGB were trained using 10-fold stratified cross validation with SMOTE applied to balance classes. Model performance was evaluated using accuracy, precision, recall, F1 score, Area Under the Receiver Operating Characteristic Curve (AUROC), and Area Under the Precision-Recall Curve (AUPRC).. Among the evaluated machine learning models, CART achieved the best overall performance for the Lasso-selected features (Accuracy = 0.77, F1-score = 0.69, AUROC = 0.74, AUPRC = 0.54), while GBM performed best for the Chi-square-selected features (Accuracy = 0.80, F1-score = 0.71, AUROC = 0.79, AUPRC = 0.64) and Boruta-selected features (Accuracy = 0.75, F1-score = 0.64, AUROC = 0.69). Overall, the GBM model with Chi-square-selected features demonstrated the strongest predictive performance, achieving the highest AUROC (0.79) and F1-score (0.71). Key predictors of high-risk age at childbirth included maternal age, early marriage, current contraceptive use, husband occupation, number of children, husband age and education, respondent age, and regional disparities. Findings were consistent with prior South Asian and LMIC studies while highlighting the added relevance of partner education and media exposure. Machine learning approaches identified the most influential socio demographic and behavioral determinants of high-risk age at childbirth in Bangladesh. The results provide an evidence-based framework to guide policymakers in designing interventions such as delaying age at first birth, expanding female education, and targeting high risk districts to improve maternal and child health outcomes.

Two Poly(methyl methacrylate) reactors were fabricated, and the flow of synthetic Dissolved Organic Carbon (DOC) was maintained. Humic acid was used as the source of organic carbon, with a water pH of 7-8.5, TDS of 220-240 mg/L, and EC of 0.4-0.5 mS/cm. Iron nano-particles were synthesized via the sol-gel method and characterized by Energy-Dispersive Spectroscopy (EDS), Dynamic Light Scattering (DLS), Fourier Transform Infrared Spectroscopy (FT-IR), and Field Emission Scanning Electron Microscopy (FE-SEM). The reactors were operated sequentially in three distinct phases: physicochemicals, biological, and bio-electrochemicals, with a direct current applied under flow-through conditions. The biological samples were stained with acridine orange (AO) and examined using CLSM. The hydraulic retention time (HRT) of each column was approximately 293 min (4.9 h), the effective working volume was 440 mL and the operating flow rate was 1.5 mL min ⁻ ¹. The Fe-FAC reactor demonstrated superior DOC removal, reducing concentrations from 3.61 ± 1.68 mg/L to 1.64 ± 1.34 mg/L, compared to the GAC column's reduction (4.26 ± 3.21 mg/L to 2.56 ± 1.08 mg/L). Flow cytometry (FCM) analysis revealed that Fe-GAC exhibited a lower event density than GAC at comparable dilution levels, suggesting a regulatory effect on microbial populations and biofilm development. In Phase I, both reactors demonstrated high DOC removal efficiencies. During phase II, removal efficiency declined in both systems. By Phase III, GAC exhibited a marked decline in performance, accompanied by the release of DOC, while Fe-GAC maintained a superior and more stable removal efficiency.

Animal depopulation is a critical component of emergency planning across U.S. livestock industries. Class A water-based foam (WBF) and high-expansion nitrogen foam (N2F) have been researched to help provide the swine industry with alternative tools to perform depopulation at larger scales. At this time, there is little data on perceived advantages or barriers identified by individuals, such as operation owners and employees, who would be closely involved in the initial phase of depopulation. Our aim was to survey farm personnel to gather logistics-focused insight upon viewing WBF and N2F demonstrations. Survey questions reflected method-specific difficulties scored using a 4-point Likert scale, along with respondent demographics and free response comments. Select survey sections were analyzed using multivariable linear regression models. Other sections, demographic data, and free response comments are presented descriptively. Data from 51 WBF and 76 N2F surveys were available, with 121 responses disclosing career position. Average (± SD) respondent age was 49.9 (± 13.2) years, with 22.3 (± 16.0) years for average experience in the swine industry. Respondents consisted of 42.2% (51/121) veterinarians, 32.2% (39/121) other farm staff, 13.2% (16/121) owners of their organization, 9.1% (11/121) animal handlers, and 3.3% (4/121) classified as "other". Depopulation method was the only variable of interest associated with differences in scores for performing tasks and method implementation, with N2F perceived as more difficult than WBF (P < 0.05). Largest proportions of comments for both methods pertained to ease of use and cost. At least one respondent suggested pre-assembled kits or services for either method. The input from a wide range of respondents emphasizes the perceived importance of method practicality for emergency planning. Our findings suggest that method demonstrations alone are insufficient; future work must focus on efforts to meet the need for developing avenues for user education for both WBF and N2F.

Rural conditions in Colombia have changed in recent years, and rural households have had to redesign their livelihood strategies (LS). Cocoa cultivation has accompanied this process of LS configuration, with a differentiated role in each household. This research analyzes the incidence of cocoa cultivation in the LS of rural households in the Colombian departments of Caquetá, Huila, Meta, and Santander. A semi-structured interview and a survey were applied in 406 rural households with cocoa production, examining the types of household LS, the specific household-level assets for cocoa production, and the perception on the representation of the crop for the households were identified. Five types of rural households were identified according to LS configuration: Cocoa Farmers, Livestock-Cocoa Farmers, Coffee-Cocoa Farmers, Diversified and Off-farm-Cocoa Farmers. It was found that cocoa cultivation has a different role and conditions in the types of households according to the LS adopted, a difference that was more accentuated at the departmental level. Households of Santander have better capital endowment for cocoa production; the opposite is the case in Caquetá. We found that cocoa production has a positive relationship (p < 0.05) with the technological level of cocoa, dedication to the crop and cocoa area. The reasons that led rural households to plant cocoa were mainly territorial conditions (not the type of rural household), economic motivation and ease of management. Finally, 90% of the households indicated that the crop has a promising future, which they associate with a livelihood, a secure source of income for old age, and the fulfillment of basic needs.

Music is a complex socio-cultural construct that fascinates researchers across many fields. Understanding its historical development may inform questions of perception and cognition, and offer insight into cultural transmission, creativity, and innovation. Here we study musical features related to harmony and document how they evolved over 400 years in western classical music. We develop a variant of the center of effect algorithm to assign the most likely key to a given set of notes, representing a piece as a sequence of local keys computed bar by bar, and we define information-theoretic measures of key uncertainty, key diversity, and novelty in key transitions. We illustrate these measures with specific examples and give them a musical interpretation, showing how they can be used to study the evolution of harmony. Applying them across the corpus, we confirm several trends previously reported by musicologists and scientists, with some discrepancies during the Classical period: in particular, we find a decline in harmonic innovation during the Early Classical period followed by a steep increase in the Late Classical, for which we offer an explanation consistent with accounts by music theorists. Comparing composers directly by their key-transition vocabularies, we further find that they group by historical period, and that both the overall arrangement and the individual nearest-neighbor relations recover connections consistent with the documented historical record. Finally, we discuss the limitations of this approach for cross-cultural studies and the need for more expressive yet tractable score representations, and a large and reliable corpus, for future study.

This study investigated how people perceive smiles that follow an angry versus a neutral expression by examining gaze patterns and subjective evaluations when presented with dynamic facial expressions. In an eye-tracking experiment, participants viewed morph movie stimuli in which facial expressions changed either from anger to a smile (anger-to-smile condition) or from a neutral expression to a smile (neutral-to-smile condition), while their gaze allocation to different facial parts (eyes, nose, mouth) was recorded. Then participants evaluated the sincerity of the final smiles and how happy and angry the poser appeared to feel. Results revealed that ratings of sincerity, happiness and anger for the final smiles did not differ between conditions, whereas gaze allocation did. Participants in the anger-to-smile condition fixated longer to the eyes; however, fixations to the eye regions did not reliably predict subjective ratings. Contrary to our predictions, longer fixation on the mouth were associated with higher ratings of sincerity and happiness regardless of conditions and with lower ratings of anger in the anger-to-smile condition. Time-course analyses for gaze distribution over time indicated that gaze shifted from the nose toward the eyes and mouth over time and exploratory analyses also showed that the temporal changes in the gaze distribution in later phases of morphing changes were associated with subjective ratings. However, these associations did not reliably differ by preceding emotional contexts. These findings suggest that, when viewing dynamic smiles embedded in a prior negative context, attention to the smiling mouth is linked to more positive evaluations, potentially reducing sensitivity to earlier negative emotional cues. In the present paradigm, the preceding facial expression did not produce detectable differences in explicit evaluations; whether such temporal information contributes to authenticity inferences remains to be tested directly. Future research should examine this question using explicit authenticity judgments and richer social-context cues that motivate masking.

Reverse engineering gene regulatory networks from gene expression data is a challenging inference task. A related problem in computational systems biology is identification of signalling networks that perform particular functions, such as adaptation. Indeed, for many research questions, there is an ongoing need for efficient inference algorithms that can identify the simplest model, from among a larger set of inter-related models, that best explains empirical observations. To this end, we introduce Sparse Likelihood-free Inference using Gibbs sampling (SLInG), a Bayesian sparse likelihood-free inference method. SLInG provides an efficient sampling method for Approximate Bayesian Computation with sparsity-inducing hierarchical priors that is widely applicable for any simulation-based model discovery task. We first apply SLInG to linear sparse regression problem using a classic dataset, before focusing on applications to biochemical network model discovery. We demonstrate that SLInG can reverse engineer stochastic gene regulatory networks from single-cell data with high accuracy, outperforming state-of-the-art correlation-based methods. Furthermore, we show that SLInG can successfully identify signalling networks that execute adaptation. Sparse hierarchical Bayesian inference thus provides a versatile and powerful tool for model discovery in systems biology and beyond.

Predicting the evolution of drug resistance remains a central challenge in HIV prevention and treatment. Evolutionary trade-offs, i.e., the fitness costs of resistance, shift as drug-selective pressure changes, a complexity amplified by fluctuating drug concentrations during long-acting prophylaxis. We develop a Darwinian evolutionary game theory model of early HIV infection under bimonthly pre-exposure prophylaxis (PrEP) with long-acting cabotegravir (CAB-LA). This work introduces a mathematical framework that couples within-host viral dynamics with adaptive evolutionary processes in a time-dependent environment to study resistance evolution under long-acting PrEP. The model incorporates the dual transmission pathways of HIV: free virion spread and direct cell-to-cell transfer through virological synapses. We quantify how resistance mutations alter viral fitness across these modes. We formulate a deterministic within-host model that explicitly incorporates the dual transmission pathways and introduce continuous resistance traits governing infectivity and drug susceptibility for each pathway. T-cell parameter estimates are informed by data assimilation using acute-stage HIV infection data. CAB-LA is modeled with time-periodic pharmacokinetics-pharmacodynamics (PK/PD), yielding dynamic fitness seascapes rather than static fitness landscapes. Time-varying drug pressure introduces a trade-off between drug resistance and infectivity, driving competition among strains that favor different transmission strategies. Our results show that fluctuating drug concentrations reshape evolutionary outcomes, generating dynamic strain competition and altering the effectiveness of long-acting PrEP in blocking infection. The model reveals threshold behavior in drug robustness to mutation, identifies conditions leading to viral control, viral escape, and delayed seroconversion, and demonstrates that resistance evolution depends on both transmission pathway and initial trait configuration. These findings provide a mechanistic explanation for delayed HIV detectability observed in CAB-LA clinical trials.

Hate speech detection in low-resource and informally written languages remains a significant challenge due to the lack of annotated corpora, orthographic variability, and complex code-mixing. Roman Urdu a non-standardized variant of Urdu written in the Latin script exemplifies these linguistic hurdles. In this paper, we propose LEXF-ATT-XLM, a novel hybrid deep learning architecture that synergizes contextual language modeling with explicit domain knowledge. Our model leverages XLM-RoBERTa for deep contextual embeddings, passed through a two-layer Bidirectional Gated Recurrent Unit (BiGRU) and a multi-head attention mechanism to capture both sequential and salient linguistic patterns. Crucially, the pooled representations are fused via a learnable gating layer (Linear + tanh) before final classification. Furthermore, we integrate a domain specific Roman Urdu hate lexicon as an auxiliary regression supervision signal within a multi task learning framework to guide the model's focus. Evaluated on the RU-HSD-30K dataset using 3-fold stratified cross-validation, the proposed model achieves an average accuracy of 88.83% and a weighted F1 score of 88.83%, with fold wise weighted F1-scores of 88.58%, 89.19%, and 88.71%. Extensive ablation studies confirm that the lexicon-guided auxiliary supervision significantly enhances the model's ability to handle lexical variations, negations, and informal spelling. These findings demonstrate the robust effectiveness of our approach in addressing the unique linguistic challenges of Roman Urdu hate speech detection.

Multiplex bead-based immunoassays (MIAs) are promising tools for simultaneously detecting humoral immunity to multiple targets, potentially playing a crucial role in serosurveillance and vaccine response assessments. However, evaluation of assay performance is paramount prior to widespread use. This study presents a performance evaluation of a pan-filovirus MIA through characterization of the analytical range for the EBOV glycoprotein (GP) target and assessments of assay precision and antigen discrimination. The precision of the MIA was evaluated by comparing the detection of anti-filovirus antibodies at two independent laboratory sites: the University of Hawai'i, Honolulu (UH), and the Institut National de Recherche Biomédicale (INRB) in Kinshasa, Democratic Republic of the Congo (DRC). Forty-six samples from Yambuku, DRC, including Ebola virus Disease (EVD) survivors and close contacts, were tested at both sites. Additionally, 907 samples were tested in DRC before and after vaccination with a prophylactic EVD vaccine, ERVEBO. Results demonstrated low variability between laboratories, with intra-assay and inter-laboratory coefficients of variation below predefined thresholds for all filovirus targets included in the multiplex panel. Analyte correlation between sites was high (r2 = 0.86-0.92). Longitudinal analysis detected increased EBOV GP reactivity following vaccination, while reactivity to non-vaccine filovirus antigens remained stable, consistent with minimal cross-reactivity in a vaccinated cohort. These findings suggest that this pan-filovirus MIA produces reproducible results across distinct laboratory settings and may serve as a useful tool for comparative serologic investigations, serosurveillance, and evaluation of EBOV vaccine-associated antibody responses.

Data scarcity is a persistent challenge in medical image analysis. Synthetic data generation using deep generative models has been proposed as a potential approach to address this limitation, yet its performance in small-data settings remains poorly characterized. This study compared three class-conditional generative approaches, a conditional variational autoencoder (cVAE), a conditional shallow-decoder VAE variant (cSD-VAE), and a conditional Wasserstein GAN with gradient penalty (cWGAN-GP), against classical geometric augmentation for emphysema subtype classification on 168 CT patches (three classes: normal tissue, centrilobular emphysema, and paraseptal emphysema). Each method was evaluated at three synthetic-to-real ratios (0.5x, 1.0x, 2.0x) using patient-level 70/30 splits across 10 random seeds, with an ImageNet-pretrained ResNet18 as the downstream classifier. No individual augmentation strategy produced a statistically significant improvement in balanced accuracy over the unaugmented baseline (0.522 ± 0.067). The conditional WGAN-GP at 1.0x achieved the highest individual balanced accuracy (0.548 ± 0.068, Cohen's d = 0.47 versus baseline), but did not reach statistical significance (p = 0.084). A pre-specified ensemble combining all four augmentation methods at the 1.0x multiplier did not significantly improve balanced accuracy over baseline (0.541 ± 0.088 versus 0.522 ± 0.067; Cohen's d = 0.32, Wilcoxon p = 0.275). Neither pixel-space nor feature-space distributional fidelity was associated with downstream classification performance. Overall, no benefit was detected from class-conditional generative augmentation in this small-data, texture-driven setting. Future work should focus on improving generative modeling under small-data conditions, including task-aware objectives and pathology-constrained synthesis.