Voice analysis as a digital biomarker: A machine learning approach for automated multiple sclerosis classification.

Despite Sorghum ( Sorghum bicolor ) potential to enhance food security and livelihoods, the commercialisation of sorghum remains limited, with most production occurring under subsistence farming systems. Our study evaluates determinants of the decision to participate in the sorghum market and the intensity of participation among smallholder farmers in Western Kenya. We conducted a cross-sectional survey of 300 farming households. Using the Heckman two-step sample selection model, our study evaluated the propensity to participate in the market and the intensity of participation. Out of the 300 households sampled, 103 (34%) participated in sorghum markets. We found that education (β = 0.63, p = 0.048), hired labour (β = 1.20, p < 0.001), log land size (β = 0.62, p = 0.070), group membership (β = 1.19, p < 0.001), occupation (β = 0.86, p = 0.018), market information (β = 2.06, p < 0.001), radio ownership (β = 0.0.51, p = 0.049), mobile ownership (β = 0.65, p = 0.045), distance to market (β = -0.35, p = 0.037), and geographical location (β = 1.18, p < 0.001) significantly determined market participation level. The intensity of participation was significantly predicted by education (β = 0.135, p < 0.001), log age (β = -0.53, p = 0.043), log family size (β = -0.267, p < 0.001), hired labour (β = 0.111, p = 0.002), radio ownership (β = 0.156, p < 0.001), mobile ownership (β = 0.213, p < 0.001) and distance to market (β = -0.160, p = 0.002). Our findings suggest that enhancing access to market information and promoting group membership can increase sorghum market participation and improve commercialisation. Policies aimed at improving rural infrastructure, strengthening agricultural cooperatives, and expanding educational opportunities for farmers are crucial for enhancing market access and supporting sorghum commercialisation. • The Heckman two-step model used to assess market participation and intensity • Sorghum market participation is driven by education, labor, and land size • Access to market information and group membership enhances market participation • Larger family size negatively impacts the intensity of sorghum market participation

Despite recent policy interventions and eradication campaigns, opium poppy cultivation remains deeply embedded in the Afghan rural economy, questioning the success of the current anti-drug strategies. The objective of this study is to investigate farm households' motives to cultivate opium poppy in Eastern Afghanistan. The study draws on a survey of farm households conducted between November 2024 and January 2025 in the eastern region. The decision to cultivate poppy and the size of the land devoted to this crop have been analysed using binary logit, Tobit, and Heckman selection models relying on a total of 616 completed interviews. The analysis reveals that past eradication campaigns have not been effective. Farmers who have experienced eradication are 13.2% more inclined to cultivate poppy than their peers. Likewise, those subjected to bribery demands and better connected to the authority are also more likely to grow poppy. Further variables positively associated with the likelihood of poppy cultivation include household debts and a lack of alternative employment opportunities. Conversely, access to extension services, credit, irrigation infrastructure, and regular religious attendance are negatively associated with poppy farming. The study recommends an integrated, development-focused policy approach, emphasising financial inclusion, non-poppy related business development, public services for agriculture, and a credible administration to achieve sustainable, long-term reduction in illegal cultivation.

Purpose Choosing the right contractor is very important for the successful management of construction projects, especially when it comes to road maintenance. In order to improve the contractor for road maintenance, this study aims to create a thorough and efficient multiple-criteria decision-making (MCDM) model that goes beyond price-only assessments to include a wider range of criteria. Design/methodology/approach The study used both the Fuzzy Analytic Hierarchy Process (F-AHP) and the Technique for Order Preference by Similarity to the Ideal Solution (TOPSIS). There were six main groups of 26 sub-criteria: Contract Bid Price and Financial Capacity, Technical Capacity, Experience, Management Capacity, Safety and Environment. F-AHP was used to figure out how important each of these groups was compared to the others. TOPSIS was then used to rank possible contractors based on these weighted criteria. Findings The analysis showed that some criteria were especially important when choosing a contractor. Bid Price, Financial Performance, History of Non-Performance of Contracts, and Timely Completion of Projects were the most important factors in deciding whether a contractor was right for the job. These results show how important it is to use both financial and performance-related indicators when evaluating the appropriate contractor. Originality/value By addressing significant flaws in the conventional selection model, this study offers a fresh methodical approach to contractor evaluation. It provides a standard for enhancing the Roads Administration's procurement procedures and establishes a strong basis for upcoming studies and policy creation targeted at boosting accountability, efficacy and transparency in the execution of road maintenance.

ABSTRACT Understanding past practices of plant resource gathering is central to the interpretation of human–environment interactions in archaeological contexts. This study examines the selection and use of wild plant resources – particularly woody taxa – by past societies that inhabited central Argentina, focusing on the Austral Pampean Hills. The analysis is based on a comprehensive review of archaeological and archaeobotanical research conducted over recent decades. Although archaeobotany in the region has developed well-established research lines, such as anthracology, microremain studies, and carpology, significant gaps persist in our understanding of gathering and selection practices across both hunter-gatherer and more sedentary societies, particularly during the Early and Middle Holocene. Ethnobotanical and ethnoarchaeological evidence from the Chaco forest phytogeographic region is integrated to highlight the diversity of uses and meanings attributed to woody species, demonstrating that individual taxa may fulfil multiple social, economic, and symbolic roles. This perspective provides a framework for developing new research questions, especially when considered alongside taphonomic processes and preservation biases in different archaeological contexts. Additionally, we examine the structural and floristic variability of Chaco forest vegetation associated with archaeological sites as an initial approach to reconstructing past plant communities and forested landscapes. While acknowledging the need for further paleoenvironmental research, this study offers a baseline for future regional-scale models of human occupation, mobility, and plant resource selection that explicitly incorporate vegetation dynamics.

Optimal selection of railway alignment schemes is a critical phase in railway line design. However, existing studies primarily adopt an engineering perspective, often overlooking economic and social factors. Current decision-making models also face limitations in quantifying qualitative indicators and balancing subjective and objective weights, thereby undermining the scientific rigor and effectiveness of the evaluation process. To address these issues, this study proposes a multi-attribute optimization model for railway alignment selection based on combined weighting and the interval number model. First, the key factors influencing railway alignment are systematically identified through expert consultations and an extensive literature review. On this basis, a multi-dimensional evaluation index system comprising 20 indicators is established, encompassing four dimensions: technical feasibility, economic rationality, environmental impact, and regional coordinated development. Subjective weights are derived using the Analytic Hierarchy Process (AHP), while objective weights are calculated through an improved CRITIC method. Game theory is then employed to integrate them. To further enhance the objectivity and robustness of the evaluation, the study develops a decision-making model based on interval number distance. This model quantifies qualitative indicators using interval numbers and ranks alternative schemes by computing their distances to an ideal solution. A case study on the Jieshipu-Pingliang section of the Dingxi-Pingliang railway confirms the method's validity. The results show that the comprehensive weighting method reduces the influence of weight factor sensitivity on route selection. Furthermore, the scheme ranking derived from the interval number model closely aligns with the scheme recommended by the design institute and demonstrates stability. The research findings can provide a scientific reference for optimizing railway alignment in economically underdeveloped regions.

ABSTRACT As regulatory and stakeholder pressures intensify, firms are increasingly expected to move beyond symbolic sustainability commitments towards corporate climate activism. This concept refers to the active institutionalisation of climate‐focused mechanisms such as external assurance, board oversight and climate‐linked incentives. Given that female directors are often associated with enhanced monitoring and greater sensitivity to long‐term stakeholder risks, this study posits that gender diversity is a crucial driver of these substantive board‐level refinements. This study investigates the impact of board gender diversity on corporate climate activism among non‐financial firms in the S&amp;P 500 index. Using a panel of 4376 firm‐year observations, we investigate whether and how gender diversity at the board level influences carbon‐sensitive firms' climate activism. Our findings show that greater gender diversity on boards drives firms' climate‐focused efforts. The positive association holds when we use alternative measures such as the Blau index and the absolute number of female directors. Further analysis suggests that the presence of two or more women on corporate boards is necessary for women to exert a significant impact on climate activism, consistent with critical mass theory. These results remain robust across instrumental variable estimation, propensity score matching and the Heckman selection model. Our study contributes to the related literature by providing empirical evidence that gender diversity plays a pivotal role in shaping firms' climate‐responsible strategies and climate risk management.

Chronic kidney disease (CKD) constitutes a critical global public health challenge. Its early-stage symptoms are subtle and easily overlooked, frequently resulting in delayed diagnosis and escalated treatment expenditures. To enable timely early intervention, this study proposes a deep learning-based multimodal CKD stage prediction model, which integrates Western medical laboratory data and traditional Chinese medicine (TCM) symptom descriptions, thereby overcoming the inherent limitations of unimodal prediction approaches. For Western medical data, the synthetic minority oversampling technique (SMOTE) was employed to address class imbalance. Subsequently, a least absolute shrinkage and selection operator (LASSO) feature selection model was utilized to identify key biomarkers, including serum creatinine and serum chloride. A backpropagation neural network, enhanced with Adam optimization and regularization mechanisms, was constructed for predictive modeling. For TCM symptom texts (e.g., tongue manifestations and pulse conditions), the Google-pretrained Bidirectional Encoder Representations from Transformers (BERT) model was leveraged to learn latent semantic patterns in the textual data. Finally, a multimodal decision fusion strategy based on the attention mechanism was adopted to dynamically learn the relative importance of Western medical and TCM features in CKD staging prediction, culminating in the development of the proposed deep learning-driven multimodal CKD staging model. Validation results indicate that the proposed model outperforms classical machine learning models and unimodal models across multiple metrics, including accuracy, F1-score, area under the curve, and convergence efficiency. These findings confirm its clinical feasibility and effectiveness, providing an innovative multimodal data-driven prediction framework that synergizes the strengths of Western medical quantitative testing and TCM qualitative diagnosis.

Industry 5.0 requires manufacturing firms to redesign supplier portfolios by jointly optimizing economic performance and environmental and social responsibilities. Conventional Multi-Criteria Decision Making (MCDM) approaches often rely on static expert weights and rich ESG data, which are rarely available for small and medium-sized enterprises and sub-tier suppliers. This study proposes an Explainable Recommender System-Based Decision Support Model (XRS-DSM) that leverages operational logistics data to approximate sustainability and risk, thereby enabling pragmatic yet transparent supplier selection. The model constructs proxy indicators for carbon impact from transportation modes and for quality from defect and inspection records, and then applies K-Means clustering to group suppliers into interpretable strategic archetypes along a Sustainability–Efficiency Frontier. A weighted utility scoring mechanism with adjustable priorities allows managers to dynamically emphasize low carbon, high quality, or low cost and immediately observe the resulting re-ranking of suppliers. The XRS-DSM is evaluated on a real-world Supply Chain Logistics Dataset from Kaggle, containing multimodal transport, inventory, and performance variables for 100 SKUs in the FMCG sector. Experimental results indicate that the model can identify Pareto-efficient supplier sets that reduce defect rates and carbon footprint with only marginal increases in landed cost, thus narrowing the “Green Premium” associated with sustainable sourcing. The proposed framework offers a scalable, data-light, and interpretable tool for manufacturing decision-makers seeking to align profitability with Industry 5.0 and ESG agendas.

Purpose This study aims to examine the marketing drivers influencing medical device selection among surgeons in suburban Tier II Indian cities, with an explicit focus on the mediating role of surgeon training. The research addresses a major gap in existing research where multiple marketing and product-related drivers have not been jointly analysed, nor has the mediating role of surgeon training been explored in resource-constrained suburban Tier II Indian markets. Design/methodology/approach A cross-sectional structured survey of 394 surgeons across 40 suburban Tier II Indian cities was conducted. Using partial least squares structural equation modeling, the effects of device safety, performance, innovation, vendor reliability, peer influence and cost consideration on device selection were assessed. Surgeon training was assessed as a mediator. Reliability, validity, mediation and structural effects were assessed using a bootstrapped sample of 5,000 iterations. Findings All predictors significantly influenced device selection (β = 0.1117–0.358, p &amp;lt; 0.01). Surgeon training partially mediated the relationships between device safety, performance, innovation and peer influence, strengthening the impact of marketing-related and peer-driven influences on adoption. The model demonstrated strong predictive power (R² = 0.71) and acceptable fit (SRMR = 0.045). Research limitations/implications The cross-sectional design limits causal inference. Specialty-specific behaviors were not analyzed. Future research should investigate longitudinal adoption patterns and comparative analysis between urban Tier I and suburban Tier II cities. Practical implications Medical device marketers targeting suburban Tier II Indian cities should prioritize surgeon training initiatives as a strategic lever to increase medical device selection. Building strong vendor relationships, emphasizing safety and performance and leveraging peer endorsements can enhance device selection. Social implications Policymakers can institutionalize and facilitate accredited surgeon training programs, introduce evidence-based procurement policies, introduce vendor reliability ratings and leverage key opinion leaders in government panels to aid device adoption. Originality/value This is one of the first empirical studies to integrate marketing, technical and social drivers into a unified model of medical device selection in a developing-country context. It advances theoretical understanding by identifying surgeon training as a critical cognitive mechanism linking marketing and clinical attributes to adoption behavior. The findings offer actionable insights for medical device marketers and policymakers operating in low-resource markets.

In transmission systems, power transformers are key high-value components, and their consistent operation is fundamental for maintaining grid stability and achieving cost-effective performance. The Transformer Health Index (THI) integrates key diagnostic parameters including dissolved gas analysis, water content, oil quality indicators, and power factor providing essential insights for asset condition assessment and investment planning. In this study, THI prediction is conducted using the Random Forest algorithm, recognized in literature for its high predictive accuracy for transformer applications, in combination with data preprocessing and filtering techniques applied to transformer dataset. For the first time, to the best of our knowledge in the THI prediction literature, various wavelet families are systematically compared at the preprocessing stage to examine their influence on predictive accuracy. The results show that the Symlet-2 configuration consistently outperformed other families in both filtered and non-filtered datasets, while Coiflet-3 and Coiflet-5 achieved higher efficiency through dimensionality reduction but with an accuracy decrease of approximately 0.09–0.10 in R2 compared to Symlet-2. The findings demonstrate that the choice of wavelet family in the preprocessing phase directly impacts feature selection outcomes and model performance, offering valuable guidance for the development of high-accuracy transformer condition assessment frameworks.

Existing tile size selection approaches are tightly coupled with compiler transformation pipelines, often leading to inaccurate modeling of cache behavior and limited effectiveness for non-rectangular tile shapes. This paper presents TileMind , a decoupled analytical model that combines compile-time and runtime information for tile size selection in affine programs. It introduces a transformation-aware pre-tiling step that enables the decoupled selector to remain consistent with compiler transformations while extracting compile-time metadata. The extracted metadata is then combined with profiled runtime characteristics to construct a richer yet tractable feasible domain, within which a nonlinear objective for tile size selection is formulated. This objective is subsequently transformed into a binary product linearization problem, with its nonlinear constraints also linearized for efficient optimization. Finally, an intra-tile optimization aligns computation with data layout to enhance data reuse within tiles. Across two multi-core Intel CPUs, TileMind achieves 1.49 × (sequential) and 1.33 × (parallel) mean speedups on twenty PolyBench kernels, and 2.08–3.54 × speedups on three deep learning workloads over the state-of-the-art analytical model Pluto-tss . Compared with TVM’s latest autotuner MetaSchedule, TileMind delivers 1.35–1.46 × mean speedups while reducing tuning overhead by 2–4 orders of magnitude. While demonstrating effectiveness on selecting tile sizes for non-rectangular tile shapes and compatibility with PPCG, Pluto, and TVM, we further provide proof-of-concept results on GPUs, illustrating the potential portability of TileMind across architectures.

Long COVID is a complex condition where symptoms persist for more than 3 months after SARS-CoV-2 infection and affects an estimated 5-30% of individuals. While persistent inflammation has emerged as an important feature of this condition, it is unclear if immune responses from COVID-19 vaccination or SARS-CoV-2 re-infection exacerbate or mirror the initial inflammatory responses. We quantified 182 inflammatory and neurology-related proteins in plasma using multiplexed affinity proteomics. Plasma samples from the COVID PROFILE cohort conducted in Victoria, Australia, were collected 6-9 months after first infection, but before COVID-19 vaccination from individuals who had recovered from COVID-19 (n = 21) or from individuals with long COVID (n = 12). To establish baseline plasma profiles, protein levels were benchmarked against unvaccinated, SARS-CoV-2 naive individuals (n = 24). In addition, we performed longitudinal analysis in a subset of individuals (n = 34), where paired samples collected 2-4 weeks after a third COVID-19 vaccine dose and after SARS-CoV-2 breakthrough infection were available to assess inflammatory and neurology protein plasma levels after antigen exposure in these contexts. In this cohort Boruta feature selection and lasso regression models identified IL-20, HAGH, NAAA, CLEC10A, LXN, and MCP-1, TRAIL, G-CSF, NBL1, and CCL23 as best discriminating proteins when comparing the long COVID group to groups of either healthy or COVID-19 recovered. Notably, longitudinal analysis indicated differences in the levels of a subset of plasma proteins following primary infection compared to after COVID-19 booster vaccination and breakthrough infection within the groups. These findings suggest that there is an altered immune response outcome primarily observed in individuals with long COVID upon re-exposure.

To evaluate the impact of constraining proprietary and open large language models (LLMs) to structured outputs in processing radiology request forms (RRF). We evaluated five LLMs-two proprietary (GPT-5-Thinking, Gemini 2.5 Pro) and three open (Qwen3-235B-A22B-Thinking, gpt-oss-120b, medgemma-27b-it)-on 100 RRFs (50 computed tomography, 50 magnetic resonance imaging). Each model processed cases with and without constraints to structured outputs. Endpoints included accuracy for modality, anatomical region, contrast phase, urgency, "all correct" (all four categories correct), and "indication improved" (clarity of rewritten text). Outputs were evaluated against a reference standard defined by board-certified radiologists and compared with two radiology residents (first-year and third-year). Accuracies with 95% confidence intervals were calculated. Constraining to structured outputs had model-dependent effects: it improved Gemini 2.5 Pro (all correct: from 53.0% [43.3-62.5] to 66.0% [56.3-74.5]) but reduced GPT-5-Thinking accuracy (from 76.0% [66.8-83.3] to 53.0% [43.3-62.5]), with minimal influence on open models. Both proprietary LLMs outperformed the best open models (up to 41.0% [31.9-50.8]). All LLMs exceeded the unassisted first-year residents' performance (19.0% [12.5-27.8]). LLM assistance improved first-year residents' accuracy to 65.0% [55.3-73.6], approaching the third-year residents' performance (80.0% [71.1-86.7]), who performed comparably to the best LLMs. Across models, performance was highest for modality and anatomical region, and lowest for urgency. Indication reformulation was judged clearer in > 90% of cases across all models without hallucinations. Constraining to structured outputs exerted model-specific effects. Proprietary LLMs achieved the highest accuracy in RRF-based protocol selection and improved first-year resident performance to an experienced-resident level. LLMs may serve as valuable decision-support tools for radiology workflow. Constraining LLMs to structured outputs produced divergent, model-specific effects in radiology protocol selection-improving Gemini 2.5 Pro, reducing GPT-5-Thinking, and minimally affecting open models-highlighting the need for model-specific prompting strategies before adopting LLMs in radiology decision support. Structured output constraints affect LLM performance differently. Gemini 2.5 Pro benefits from structured prompting, while GPT-5-Thinking declines. Open-weight models show minimal impact from output constraining. Proprietary models outperform open models in radiology protocol selection.

The present study aimed to assess the link between physical appearance anxiety (PAA) and eating disorder (ED) symptomatology by a meta-analysis of existing literature. Eligible studies were searched across six electronic databases up until November 20, 2025. Pooled effect sizes (r) were calculated using random-effects models. Potential variables that influence effect heterogeneity were analyzed by univariable and multivariable meta-regressions. Influence analyses and a three-parameter selection model (3PSM) were used to assess robustness of the results and publication bias. Twenty-seven effect sizes from 21 studies (N = 5261) were obtained. The results indicated a strong association (i.e., r = 0.559) between the two variables under consideration, which was notably stronger (i) among females compared to males; and (ii) for overall eating disorder symptoms rather than bulimic symptoms. The results of this study advocate for further investigation into the effectiveness of addressing anxiety responses related to personal body traits, particularly among females, within the context of preventing and treating eating disorders.

Antibodies are encoded by nucleotide sequences that are generated by V(D)J recombination and evolve according to mutation and selection processes. Existing antibody language models, however, focus exclusively on antibodies as strings of amino acids and are fitted using standard language modeling objectives such as masked or autoregressive prediction. In this paper, we first show that fitting models using this objective implicitly incorporates nucleotide-level mutation processes as part of the protein language model, which degrades performance when predicting effects of mutations on functional properties of antibodies. To address this limitation, we devise a new framework: a deep amino acid selection model (DASM) that learns the selection effects of amino acid mutations while explicitly factoring out the nucleotide-level mutation process. By fitting selection as a separate term from the mutation process, the DASM exclusively quantifies functional effects: effects that change some aspect of the function of the antibody. This factorization leads to substantially improved performance on standard functional benchmarks. Moreover, our model is an order of magnitude smaller and multiple orders of magnitude faster to evaluate than existing approaches, as well as being readily interpretable.

Selecting appropriate wall materials for low-cost housing is a complex decision-making process involving multiple technical, economic, social, and environmental considerations. This study aims to identify and prioritize the most influential attributes affecting wall material selection for affordable housing. A total of twenty-two attributes were derived from a comprehensive literature review and evaluated through a structured questionnaire survey involving 30 respondents, consisting of academics, decision-makers, and technical practitioners. A consensus-based weighting approach was applied using the mean value (μᵢ), normalized weight (wᵢ), and relative ratio (r) to establish the priority structure of attributes. The results indicate that all attributes are considered relevant (r &gt; 0.1), with the highest priorities assigned to structural strength and stability (T1, w = 0.060), initial wall cost (E1, w = 0.058), life-cycle cost (E2, w = 0.057), durability and weather resistance (T3, w = 0.056), occupant safety and perceived security (S4, w = 0.056), and embodied carbon and energy (L1, w = 0.053). At the aspect level, technical factors contributed 34% of the total weight, followed by economic (26%), social (21%), and environmental aspects (19%). These findings provide a quantitative attribute-weighting framework that can support the development of Multi-Attribute Decision Making (MADM) models for context-sensitive wall material selection in low-cost housing.

Aptamers in cancer therapy: Why has clinical translation lagged behind preclinical promise?

Multi-objective decision model for wastewater treatment technology selection based on machine learning.

BDGraS: Bandwidth-adaptive dual-relation gravity model for efficient cooperative vehicle selection in autonomous driving