Heliyon

Road traffic accidents represent a global public safety crisis, necessitating advanced computational tools for accurate injury severity prediction and effective decision support. This study evaluates high-performing ensemble machine learning models, including AdaBoost, XGBoost, LightGBM, HistGBRT, CatBoost, Gradient Boosting, NGBoost, and Random Forest, using a comprehensive National Highway Traffic Safety Administration (NHTSA) dataset from 2018 to 2022. While all models demonstrated exceptional predictive accuracy, with HistGBRT achieving the highest overall accuracy of 92.26%, a defining achievement of this work is the perfect classification (100% precision and recall) of fatal injuries across all ensemble architectures. To bridge the gap between predictive performance and actionable intelligence, this research integrates SHapley Additive exPlanations (SHAP) to provide both global insights into dataset-wide risk factors and local, instance-specific rationales for individual crash events. The global analysis identified ethnicity, airbag deployment, and harmful event type as primary drivers of injury severity, while local force and waterfall plots revealed the precise "push and pull" of variables for specific incidents. The results offer a robust, interpretable framework for stakeholders tasked with improving traffic safety and mitigating crash-related harm.

Despite the remarkable progress of artificial intelligence represented by large language models, how AI technologies can contribute to the construction of evidence in evidence-based medicine (EBM) remains an overlooked issue. Now, we need an AI that can be compatible with EBM. In the present paper, we aim to propose an example analysis that may contribute to this approach using variable Vision Transformer.

ObjectiveThis study evaluates the impact of systemic GLP-1 receptor agonist (GLP-1RA) use on surgical wound healing in high-risk surgical populations, including patients with diabetes, and implications for perioperative planning and healing outcomes. ApproachThis pilot retrospective cohort study compared adult surgery patients with non-healing postoperative wounds by their GLP-1RA use. Outcomes included healing status, time to wound closure, and number of surgical interventions. ResultsThe cohort included 35 non-GLP-1RA users and 16 GLP-1RA users with comparable baseline characteristics, except for significant higher prevalence of venous insufficiency among users. Though median time to closure was similar for all patients, users required fewer surgical interventions and their wounds reached closure in significant difference from non-users. Among patients with diabetes, all GLP-1RA users healed significantly compared to non-users. InnovationThe impact of GLP-1RA therapy on wound healing in high-risk reconstructive and soft-tissue surgery remains poorly defined. This pilot cohort addresses that gap, offering an early signal that GLP-1RA use is associated with improved wound healing and fewer postoperative interventions. These findings may inform perioperative practice by identifying a systemic pharmacologic factor that optimizes surgical outcomes in high-risk populations. ConclusionGLP-1RA use was associated with higher healing rates and fewer interventions, particularly among patients with diabetes. These findings support a beneficial role in surgical wound healing and warrant larger multi-site studies.

Machine learning models have seen a lot of success in medical image segmentation domain. However, one of the challenges that they face are confounders or shortcuts: spurious correlations or biases in the training data that affect the resulting models. One example of such confounders for surgical machine learning is the setup of surgical equipment, including tools and lighting. Using the task of identification of safe and dangerous zones of dissection in laparoscopic cholecystectomy images and videos as a use-case, we inspect two equipment-induced biases: the presence of surgical tools in the field of view and the position of lighting. We propose methods for evaluating the severity of these biases and augmentation-based methods for mitigating them. We show that our tool bias mitigations improve the models' consistency under tool movements by 9 percentage points in the most inconsistent cases, and by 4 percentage points on average. Our lighting bias mitigations help reduce fraction of true dangerous zone pixels that may be predicted as safe under light changes from 5% to 1.5%, without compromising segmentation quality.

The molecular basis of triple-negative breast cancer (TNBC), a highly aggressive and therapy-resistant subtype of breast cancer, is poorly understood. This study aims to identify key genes and pathways involved in TNBC development and progression using a systems biology approach followed by experimental validation. Here, two transcriptome microarray datasets from the GEO database were analysed using the R package LIMMA to detect differentially expressed genes (DEGs) in TNBC tumors. Gene Ontology (GO) and Kyoto Encyclopaedia of Genes and Genomes (KEGG) enrichment analyses using the DAVID database were performed to identify DEGs regulated biological functions and pathways. Further, a protein-protein interaction (PPI) network was constructed using the STRING online database, and the topological properties were determined using MCODE and Cytohubba plug-ins. The expression and the prognostic value of the hub genes were validated using the Cancer Genome Atlas (TCGA) survival analysis. We found 727 DEGs, of which 473 were downregulated and 254 were upregulated in TNBC vs. non-TNBC samples. The GO and KEGG analyses indicated that the DEGs were mainly related to cell adhesion, tumorigenesis, and cellular immunity. The PPI network had shown six hub genes, namely CCND1, CDH1, ESR1, FN1, IL6, and PPARG, as the top key regulators. All these genes were validated by quantitative real-time PCR in the TNBC cell line using non-TNBC cell line as a calibrator, and the obtained results were in accordance with the bioinformatics data. This information may contribute to understanding the various molecular mechanisms that drive the development and progression of TNBC tumors.

ObjectiveThe purpose of the present study is to describe the survival outcomes of patients with low-grade serous ovarian cancer (LGSOC) in the post-operative setting from a tertiary gynecologic oncology referral centre in Quebec, including evaluation of patient characteristics, clinical outcomes and prognostic factors. MethodsThe study included 25 patients: 1) with a post-surgical histopathologic diagnosis of a low-grade serous tumour of the ovary, 2) underwent primary cytoreductive surgery prior to adjuvant therapy, and 3) for whom clinical data was available. Clinical and demographic features were characterized by descriptive statistics. Clinical endpoints of progression-free survival (PFS) and overall survival (OS) were assessed, utilizing the Kaplan-Meier method for estimating survival probabilities. ResultsThe median age of this cohort was 61 years (range, 26-81). Median OS was 140.6 months in patients with no residual disease (R0), 71 months in patients with microscopic residual disease (R1), and 27.7 months in patients with macroscopic residual disease (R2) (p=.001). Residual disease was also found to significantly impact PFS (p=.008). Administration of adjuvant chemotherapy failed to improve survival outcomes altogether (PFS, p = .270; OS, p = .300). ConclusionsThis study supports the shifting consensus that optimal cytoreductive surgery, where feasible, is paramount for successful treatment of LGSOC. Furthermore, treatment with adjuvant chemotherapy may lead to worse survival outcomes.

There is an ongoing Oropouche Fever (OF) outbreak in Brazil since 2024. There are dengue and chikungunya prediction models available, but none to help discriminate dengue, chikungunya, and OF. Objective: This study aims to develop and validate clinical prediction models for dengue, chikungunya, OF. Methods: This study uses surveillance data from Espirito Santo state / Brazil, from 2023-2025. Epidemiological investigations and biological samples were used to conclude cases as either (a) clinical-epidemiologically confirmed, (b) laboratory confirmed, or (c) discarded. The predictors were all data related to signs, symptoms, and comorbidities available in the notification forms. The analysis was performed using random forest regression models, one for each outcome, in development and validation datasets. Results: A total of 465,280 observations were analyzed, 261,691 dengue cases (56.6%), 18,676 chikungunya cases (4.0%), 12,174 OF cases (2.6%), and 179,115 discarded cases (38.6%). All three models had good discrimination and moderate to good calibration after scaling prediction. The models retained from 26 to 16 predictors each. Leukopenia and vomiting were the most discriminatory predictors for dengue, arthritis, arthralgia, and rash were the most discriminatory for chikungunya, and epidemiological features were the most relevant for OF. The dengue, chikungunya, and OF models had ROC AUC of 0.726, 0.851, and 0.896 in the validation set, respectively. Conclusion: This research identified predictors most discriminative between dengue, chikungunya, and OF. We developed and validated predictive models, one for each condition, with moderate to very good performance available at https://pedrobrasil.shinyapps.io/INDWELL/. One may use them in diagnostic work-up and arbovirus surveillance.

Understanding the dynamics of HIV epidemics is important to control them effectively. Classical methods that mainly rely on occurrence data are limited by the fact that an unknown part of the epidemic eludes sampling. Since the early 2000s, phylodynamic methods have enabled the estimation of key epidemiological parameters from virus genetic sequence data. These methods have the advantage of being less sensitive to partial sampling and to provide insights about epidemic history that even predates the first samples. In this study, we analysed 2,205 HIV sequences from the French ANRS PRIMO C06 cohort. We identified and were able to reconstruct the temporal dynamics of two large clades that represent the HIV-1 epidemics in the country. Using Bayesian phylodynamic inference models, we found that the first clade, from subtype B, originated in the end of 1970s, grew rapidly during the 80s before decreasing from 2000 to 2015 and stagnating since then. The second clade, from circulating recombinant form CRF02_AG, emerged and spread in the 80s, grew again in the early 2000s, before declining slightly. We also estimated key epidemiological parameters associated with each clade. Finally, using numerical simulations, we investigated prospective scenarios and assessed the possibility to meet the 2030 UNAIDS targets. This is one of the rare studies to analyse the HIV epidemic in France using molecular epidemiology methods. It highlights the value of routine HIV sequence data for studying past epidemic trends or designing public health policies.

The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has continuously evolved since its emergence in the human population in 2019. As of 1st August 2025, more than 1,700 Omicron subvariants have been designated by the Pango nomenclature system. The Pango nomenclature system designates a new lineage based on genetic and epidemiological information of SARS-CoV-2 strains. However, there is a possibility that strains that have similar genetic backgrounds and the same phenotype are given different Pango lineage names. In this paper, we propose a new algorithm, called FindPart-w, which can identify groups of viral lineages that share the same relative effective reproduction numbers. We introduced a new lineage replacement model, called the constrained RelRe model, which constrains groups of lineages to have the same relative effective reproduction numbers. The FindPart-w algorithm searches the equality constraints that minimise the Akaike Information Criterion of constrained RelRe models. Using hypothetical observation count data created by simulation, we found that the FindPart-w algorithm can identify groups of lineages having the same relative effective reproduction number in a practical computational time. Applying FindPart-w to actual real-world data of time-stamped lineage counts from the United States, we found that the Pango lineage nomenclature system may have given different lineage names to SARS-CoV-2 strains even if they have the same relative effective reproduction number and similar genetic backgrounds. In conclusion, this study showed that viruses that had the same relative effective reproduction number were identifiable from temporal count data of viral sequences. These findings will contribute to the future development of lineage designation systems that consider both genetic backgrounds and transmissibilities of lineages.

Background: Elevated lipoprotein(a) [Lp(a)] is a known risk factor for several cardiovascular-related diseases established from multiple genetic and observational studies. However, the underlying mechanisms mediating the effects of Lp(a) levels on cardiovascular disease risk and major adverse cardiovascular events (MACE) are unclear. The aim of this study was to identify proteins downstream of Lp(a) using mendelian randomization (MR) - a genetic causal inference approach. Methods: A two-sample MR was performed by initially identifying Lp(a) genetic instruments based on data from genome wide association studies (GWAS) of Lp(a) blood concentrations. These instruments were then tested for association with proteins from proteomic pQTL data (Olink from UK Biobank, 2940 proteins and SomaScan from deCODE, 4907 proteins). Results: A total of 521 proteins associated with Lp(a) were identified. Using pathway enrichment analysis, the following MACE-relevant pathways were identified comprising a total of 91 Lp(a) downstream proteins: oxidized phospholipid-related, chemotaxis of immune cells and endothelial cell activation, pro-inflammatory monocyte activation, neutrophil activity, coagulation, and lipid metabolism. Conclusion: The results suggest that the influence of Lp(a) treatments is primarily through modifying inflammation rather than lipid-lowering, thus providing insight into the mechanistic framework which mediates the effects of elevated Lp(a) on atherosclerotic cardiovascular disease.

Background: Medical imaging, especially computed tomography and magnetic resonance imaging, is essential in clinical care of patients with renal cell carcinoma (RCC). Artificial intelligence (AI) research into computer-aided diagnosis, staging and treatment planning needs curated and annotated datasets. Across literature, The Cancer Genome Atlas (TCGA) datasets are widely used for model training and validation. However, re-annotation is often necessary due to limited access to public annotations, raising entry barriers and hindering comparison with prior work. Methods: We screened 1915 CT scans from three TCGA-RCC databases and employed a segmentation model to annotate kidney lesion. After a meta-data-based exclusion step, we hosted a reader study with all papillary (n=56), chromophobe (n=27) and 200 randomly selected clear cell RCC cases. Two students quality checked and corrected the data as well as annotated tumors and cysts. Uncertain cases were checked by a board-certified radiologist. Results: After data exclusion and quality control a total of 142 annotated CT scans from 101 patients (26 female, 75 male, mean age 56 years) remained. This includes 95 CTs with clear cell RCC, 29 with papillary RCC and 18 with chromophobe RCC. Images and voxel-level annotations of kidneys and lesions are open sourced at https://zenodo.org/records/19630298. Conclusion: By making the annotations open-source, we encourage accessible and reproducible AI research for renal cell carcinoma. We invite other researchers who have previously annotated any of these cohorts to share their annotations.

Background: Sub-Saharan Africa (SSA) still experiences a high burden of micronutrient deficiencies. For monitoring of micronutrient status among young children in SSA, non-invasive alternatives to blood-based biomarkers are desirable. Handheld Raman spectrophotometry appears to offer this alternative to quantify intracellular stores of micronutrients. In rural Burkina Faso and Kenya, we validated the Cell-/SO-Check device (ZellCheck(R)) against conventional laboratory-based methods. Methods: For this validation study, we recruited children aged [≥]24 months attending routine clinics within the Health and Demographic Surveillance Systems (HDSS) in Siaya and Nouna. Anthropometric measurements and venous blood samples were taken. Plasma ferritin, soluble transferrin receptor (sTfR) and C-reactive protein (CRP) were measured by ELISA, and plasma zinc by atom absorption. The spectrometer was used to quantify zinc and iron. For continuous outcomes, we generated Bland Altman plots and calculated bias and limits of agreement (LoA). For binary outcomes, we produced Receiver Operator Characteristic (ROC) areas under the curve (AUC), and estimated sensitivity, specificity and predictive values. Results: We analysed data of 48 children from Burkina Faso and 54 children from Kenya (male: 53%; age range: 24-66 months). According to spectrophotometry, the proportions of iron deficiency and zinc deficiency were 16.7% and 25.5%, respectively. The median concentrations were for ferritin 24.0 {micro}g/L (range: 2.0-330.0), for sTfR 5.7 mg/L (2.8-51.0), and for zinc 9.9 {micro}mol/L (5.2-25.0). The corresponding bias for iron levels by spectrophotometry was 42.4 with LoA: -18.7, 103.6. The bias for zinc levels was 7.5 with LoA: -49.3, 64.2. For the classification of deficiency, the ROC-AUC, sensitivity, and specificity for spectrophotometry vs. biomarker-based diagnosis were for iron deficiency 0.62, 68% and 55%, respectively, and for zinc deficiency 0.55, 33% and 91%, respectively. Conclusions: The Cell-/SO-Check device may be used to rank children in population-based studies in SSA according to their zinc status, but not iron status. The method should not replace the standard laboratory measurements for clinical diagnoses of zinc and iron deficiencies.

Introduction: Accurate stratification of hard atherosclerotic cardiovascular disease (ASCVD) risk remains challenging despite advances in prevention. Liver function biomarkers (LFBs), particularly gamma - glutamyl transferase (GGT), have been linked to cardiovascular outcomes, yet their contribution to hard ASCVD risk prediction is not well defined. Methods: This study analyzed data from the National Health and Nutrition Examination Survey (NHANES, 2005 - 2018) to assess cross - sectional associations between LFBs and 10 - year hard ASCVD risk estimated by the ACC/AHA Pooled Cohort Equations. Multivariable regression, restricted cubic splines, and mediation analyses were applied to examine independent and dose - response relationships. External validation was performed in the China Health and Retirement Longitudinal Study (CHARLS) and NHANES using machine learning models (CoxBoost, Naive Bayes and Random Forest). Results: Among 5,731 NHANES participants, GGT showed an independent linear association with hard ASCVD risk (P - trend = 0.003), partly mediated by systolic blood pressure (44.8%), HbA1c (19.0%), and high density lipoprotein cholesterol (13.4%). Machine learning (ML) models incorporating GGT, alkaline phosphatase (ALP), and globulin alongside traditional risk factors improved predictive accuracy, with Naive Bayes achieving an AUC of 0.751 in NHANES validation. Conclusions: GGT is an independent and biologically plausible biomarker of hard ASCVD risk, acting through cardiometabolic pathways. Incorporating LFBs into risk prediction models, particularly with machine learning, enhances risk stratification and may facilitate early identification of high - risk individuals.

Introduction Large language models are increasingly being used in healthcare. In interventional pain medicine, clinical reasoning is essential for procedural planning. Prior studies show that simplified prompts reduce clinical detail in AI-generated responses. It remains unclear whether this reflects knowledge loss or simply prompt-driven suppression of information. Methods We performed a controlled comparative study using 15 standardized low back pain questions representing common interventional pain questions. Each question was submitted to ChatGPT under three conditions, professional-level prompt (DP), fourth-grade reading-level prompt (D4), and clinician-directed rewriting of the D4 response to a medical level (U4[->]MD). No follow-up prompting was allowed. Three physicians independently rated responses for accuracy using a 0-2 ordinal scale. Clinical completeness was determined by consensus. Word count and Flesch-Kincaid Grade Level (FKGL) were also measured. Paired t-tests compared conditions. Results Accuracy was highest with professional prompting (1.76). Accuracy declined with the fourth-grade prompt (1.33; p = 0.00086). When simplified responses were rewritten for clinicians, accuracy returned to baseline (1.76; p {approx} 1.00 vs DP). Clinical completeness followed the same pattern showing DP 80.0%, D4 6.7%, U4[->]MD 73.3%. Fourth-grade responses were shorter and less complex. Upscaled responses were more complex and similar in length to professional responses. Inter-rater reliability was low (Fleiss {kappa} = 0.17), but trends were consistent across conditions. Conclusions Reduced clinical detail under simplified prompts appears to reflect constrained output rather than loss of knowledge. Clinician-directed reframing restores omitted content. LLM performance in interventional pain depends strongly on prompt design and intended audience.

We are pleased to submit our Original article entitled "Assessing medication-related burden and medication adherence among older patients from Central Nepal: A machine learning approach" for consideration in your esteemed journal. In this paper, we assessed medication burden using validated Living with medicines Questionnaire (LMQ-3) and medication adherence using Adherence to Medication refills (ARMS) Scale. In this paper we analysed our result through machine learning approach in spite of traditional statistical approach to identify the complex factors influencing both. Six ML architectures (Ordinary Least Square, LightGBM, Random Forest, XGBoost, SVM, and Penalized linear regression) were employed to predict ARMS and LMQ scores using various socio-demographic, clinical and medication-related predictive features. Model explainability was provided through SHAP (Shapley Additive exPlanations). Our study identified the moderate medication burden with moderate non-adherence among older adults. Requiring assistance for medication and polypharmacy were the strongest drivers for the medication burden and non-adherence. The high predictive accuracy by ML suggests the appropriate clinical intervention like deprescribing to cope with the high prevalent medication burden and non-adherence among older adults in Nepal.

Research in cognitive neuroscience has relied on simple, highly controlled stimuli due to the difficulty in developing standardized, ecologically valid stimulus sets. However, there is a consensus that using ecologically valid stimuli is imperative to generalize results beyond controlled laboratory settings. The current study introduces a naturalistic audio stimulus database, consisting of short, recognizable, and emotionally rated stimuli. To create such a database, the current study collected 291 audio files from a wide range of sources. 361 participants rated the audio clips on emotionality, arousal, and recognizability, and subsequently freely described the audios by typing what they believed the sound to be. The text responses of the participants were embedded and clustered using an unsupervised machine-learning algorithm to derive a participant-grounded organization of auditory object categories. The results indicate audio clips were easily recognizable, while emotionality and arousal ratings showed broad variability, making the database suitable for diverse experimental needs. Furthermore, the final database comprises 10 distinct semantic categories, providing a diverse set of auditory stimuli.

Sequencing the respiratory tract transcriptome has the potential to provide insights into infectious pathogens and the hosts immune response. While DNA-based sequencing is more standard in clinical laboratories due to its stability, RNA assays offer unique advantages. RNA reflects dynamic physiological changes, and for RNA viruses, viral RNA particles directly represent copies of the viral genome, enabling greater diagnostic sensitivity. However, RNAs susceptibility to degradation remains a significant challenge, particularly in RNase-rich specimens like saliva. To address this, we conducted a systematic, combinatorial evaluation of 24 distinct mNGS workflows, crossing eight nucleic acid extraction methods with three RNA-Seq library preparation protocols. Remnant saliva samples (n = 6) were pooled and spiked with MS2 phage as a control. The SARS-CoV-2 virus was spiked into half of the samples, which were extracted using the eight different extraction methods (n = 3) and compared using RNA Integrity Number equivalent (RINe) scores and RNA concentration. The extracted RNA was then processed across the three library construction methods and subjected to short-read sequencing to assess all 24 combinations head-to-head. We compared methods based on viral read recovery and found that RINe and concentration did not correlate with viral detection. The Zymo Quick-RNA Magbead kit and the Tecan Revelo RNA-Seq High-Sensitivity RNA library kit were the extraction and library-preparation kits that yielded the most SARS-CoV-2 reads, respectively. Importantly, our combinatorial analysis revealed that any small variability attributable to different nucleic acid extraction methods was heavily overshadowed by differences in quality attributable to the RNA-Seq library preparation methods. These findings challenge the reliance on conventional RNA quality metrics for clinical metagenomics and underscore the need to redefine extraction quality standards for mNGS applications. IMPORTANCEmNGS is a powerful and unbiased approach towards pathogen detection that has mostly been applied to blood and cerebrospinal fluid samples. However mNGS has recently been applied to more areas including the respiratory pathogen detection space, with potential applications in both in-patient diagnostics and public health surveillance. Saliva samples are an ideal sample type for these use cases since they can be collected non-invasively. However, saliva is also a challenging sample type due to its high RNase activity and often yields low-quality nucleic acid. This study explores the feasibility of using saliva specimens in mNGS with contrived SARS-CoV-2 samples to optimize the combination of two factors: nucleic acid extraction and RNA-seq library preparation. Exploration in this area could enhance the sensitivity of saliva-based mNGS assays, with the goal of future expansion of this specimen type in clinical diagnostics and public health surveillance. Key PointsO_LIThe choice of RNA-Seq library preparation kit has a greater impact on pathogen detection than the nucleic acid extraction method. C_LIO_LIThe combination of Zymo Quick-RNA Magbead extraction kit and TECAN Revelo RNA-Seq High Sensitivity RNA library kit recovered the highest percentage of total SARS-CoV-2 reads. C_LIO_LIRNA quantity and RINe score do not correlate with viral read capture, indicating a need for an alternative metric to assess RNA quality for downstream mNGS clinical diagnostics. C_LI

Foodborne illness is a significant public health concern worldwide. Shiga toxin-producing Escherichia coli and Campylobacter species are recognized as important zoonotic bacterial pathogens contributing to human infections through the food chain, particularly via foods of animal origin. Although goat meat is in high demand in the southern region of Thailand, studies on foodborne pathogens in this livestock species remain limited. The current study aimed to (i) determine the antimicrobial susceptibility of Campylobacter spp. and STEC isolated from goats, and (ii) analyze the genetic relationships among Campylobacter spp. And E. coli O157 isolates obtained from different sources. Campylobacter jejuni and C. coli isolates were characterized based on sequences of seven housekeeping genes using the Achtman multilocus sequence typing scheme. For E. coli O157:H7, core genome multilocus sequence typing analysis was performed using whole-genome sequencing data. Genetic diversity was observed among C. jejuni, whereas a clonal population structure was detected in C. coli and E. coli O157:H7. Overlapping genetic characteristics were observed between C. jejuni isolates from goats and those previously reported in livestock and humans in Thailand. Among Campylobacter species, resistance to fluoroquinolones, including ciprofloxacin and nalidixic acid, was observed, whereas resistance to fosfomycin was most frequently detected in Shiga toxin-producing E. coli. Tetracycline-resistant isolates were identified in both Campylobacter species and Shiga toxin-producing E. coli at moderate levels. A multidrug-resistant pattern was observed only in C. coli, whereas no multidrug-resistant C. jejuni or Shiga toxin-producing E. coli isolates were detected. These findings indicate that healthy goats may serve as potential reservoirs of zoonotic pathogens and antimicrobial resistance in southern Thailand, where goat meat is frequently consumed.

Background: Sleep wake and circadian disturbances are increasingly recognised in people living with amyotrophic lateral sclerosis (plwALS), but endogenous circadian phase timing and its prognostic significance in early disease remain unclear. We assessed whether salivary dim-light melatonin onset (DLMO), an objective marker of central circadian phase, is altered in early plwALS and whether it provides prognostic information. Methods: In this prospective longitudinal observational study, plwALS within 18 months of symptom onset underwent home-based salivary melatonin sampling under dim light conditions at six predefined time points around habitual sleep onset (HSO). Melatonin profiles were modeled using cubic smoothing splines, and DLMO was defined as the first time the fitted curve reached 3 pg/mL. Clinical, respiratory, and sleep assessments were collected at baseline (T0) and after 6 months (T6); a subgroup repeated saliva sampling at T6. Age and sex matched controls underwent melatonin profiling. Associations with disease progression, incident respiratory symptoms, and survival/tracheostomy were examined using regressions and survival analyses. Results: Fifty plwALS were enrolled. Compared with controls, plwALS showed an earlier DLMO (20:24 vs 20:58; p=0.028) despite similar HSO and chronotype. Within ALS cohort, a later baseline DLMO correlated with worse functional/motor status, faster progression of disease, incident dyspnea/orthopnea by T6 (adjusted OR 3.02; p=0.017), and poorer survival/tracheostomy-free outcome. In re-sampled subgroup (n=28), DLMO and other melatonin-derived metrics did not change over 6 months. Conclusions: Circadian phase alterations are detectable in early ALS. Baseline DLMO may represent a non-invasive prognostic biomarker for progression, respiratory symptom emergence and survival, warranting validation in larger multicentre cohorts.

Patients with amyotrophic lateral sclerosis (ALS) face critical decisions regarding life-sustaining treatments, such as invasive mechanical ventilation and percutaneous endoscopic gastrostomy. Advance care planning and shared decision-making are standard supportive frameworks but they often fail to account for structural pressures like progressive decline, shifting patient values, and fear of becoming a burden that may influence decision-making. This study explores how patients with ALS interpret ventilator and care options amid progressive physical decline, thereby reconsidering approaches to decision support. Using a qualitative descriptive design, the researcher (a nurse/sociologist) conducted 2-3 hour home interviews with five purposively sampled patients with ALS. Data, including eye-tracking-aided responses, were analysed via Sandelowskis framework. Rigour was ensured through team-based triangulation, independent coding by two researchers, and a reflexive audit trail. Subjective narratives were prioritised without medical record cross-referencing to capture patients experiences. Four categories emerged: (1) Rewriting clinical prognosis into a narrative of exploration via peer models, where meeting active ventilator users transformed future perceptions; (2) The conflict between securing care infrastructure and the burden on family, which greatly influenced the will to survive; (3) Existential fluctuation, where patients intentions shifted with daily fulfilment and family events; and (4) Governance of the body via pre-emptive technology use and training carers as physical extensions. Findings showed decision-making was a multi-layered process redefining lifes meaning within social resources. This necessitate shifting from independent to relational autonomy, where agency relies on care infrastructure, not physical ability. Treatment choice is a dynamic exploration requiring narrative companions to support existential fluctuations. Professionals must coordinate environments to reduce patient indebtedness. Limitations include the small, resource-advantaged sample (N = 5) and reliance on subjective narratives without medical record verification. Living with ALS means governing a new self through relational support and continuous dialogue.