38978 research outputs found
Sort by
Digital Era in Full Arch Rehabilitation: A Scoping review of Trends, Applicability, and Implications for Clinical Practice
No full textThis scoping review aimed to map the available evidence on the clinical applicability and implications of digital workflow in various interventions for the rehabilitation of fixed and removable full-arch edentulous patients, with a focus on workflow-related operational outcomes and clinician-reported measures
When Should Language Models Remain Silent? Governing LLM Behavior Through a Control-Layer Approach
No full textThis project hosts the preprint:
“When Should Language Models Remain Silent? Governing LLM Behavior Through a Control-Layer Approach”
Canonical, citable version (DOI):
https://doi.org/10.5281/zenodo.18737646
Abstract:
Large Language Models (LLMs) have rapidly become central components of interactive systems for learning, problem solving, coding assistance, and decision support. In recent years, advances in model architecture, scale, and training data have substantially improved linguistic fluency and reasoning capabilities, enabling LLMs to respond accurately and helpfully to a wide range of tasks. Consequently, much of the current research and development has focused on what language models should know and how they should generate responses, often emphasizing the accuracy, reasoning depth, and alignment of the generated content.
However, as LLMs transition from static tools to continuous interactive partners, a different class of problems has begun to surface that is largely orthogonal to model intelligence. In real-world interactions, the primary failure mode is often not incorrect responses but rather poorly timed interventions. Prior work in human–computer interaction has shown that interruptions and poorly timed assistance can disrupt cognitive flow and increase cognitive load, even when the assistance itself is correct.
This study introduces a model-agnostic control-layer framework that governs response timing and verbosity in LLM-based systems, treating silence, delay, and abstention as intentional system actions
Execution-Bound Enforcement for AI Agent Governance
No full textThis project presents an architectural framework for execution-bound governance of AI agents.
The paper introduces execution-bound enforcement as a structural component of AI governance, ensuring that authority and admissibility are validated deterministically at execution time.
The framework complements identity-centric authorization approaches by providing execution-time enforcement and admissibility evidence mechanisms.
This work was prepared as a public comment submission for the NIST NCCoE Software and AI Agent Identity and Authorization Concept Paper (2026)
Patient-reported HRQoL after out-of-hospital and in-hospital cardiac arrest: a scoping review of measures and follow-up schedules
No full textStudy Information
Title
Patient-reported HRQoL after out-of-hospital and in-hospital cardiac arrest: a scoping review of measures and follow-up windows.
Rationale
HRQoL measurement after cardiac arrest is inconsistent in both instrument choice and timing, which limits comparability and the practical translation of survivorship guidance into follow-up pathways. This scoping review maps which PROMs are used and when, across OHCA and IHCA survivorship research.
Aims
Identify PROMs used to measure HRQoL (and closely allied “health status” measures when presented as HRQoL).
Map follow-up windows/timepoints used after cardiac arrest.
Describe heterogeneity by setting (OHCA/IHCA), era, and study type.
Hypotheses
No confirmatory hypotheses (scoping review).
Prespecified expectations (for structured mapping, not hypothesis testing):
A few generic instruments will dominate (e.g., EQ-5D family, SF family, PROMIS/WHOQOL).
Timepoints will cluster around common windows (e.g., ~3/6/12 months), with substantial between-study variation.
OHCA/IHCA may differ in practices, but reporting limitations may reduce comparability.
Design Plan
Study type
Scoping review of published and grey literature.
Study design
Scoping review with database searching + grey literature retrieval + citation chasing as feasible, reported using PRISMA-ScR.
Blinding
Not applicable (no intervention assignment).
Is there any additional blinding in this study?
Not applicable.
Randomization
Not applicable (no participant allocation). Any “random sampling” would only be used for internal QA (e.g., calibration checks).
Sampling Plan
Existing Data
Yes — bibliographic records, full texts where accessible, and grey literature outputs.
Registration prior to accessing the data
This preregistration is completed prior to formal screening and extraction. Searches may have already been run as part of scoping (common in reviews), but inclusion decisions and extraction follow the preregistered plan.
Explanation of existing data
Data sources include:
PubMed Query A (high recall) + Query B (instrument-filter)
search_strategies
CINAHL (EBSCO) Query A (HRQoL) + Query B (HRQoL + survivorship/time + instrument filter)
search_strategies
Web of Science combined boolean strategy
search_strategies
Grey literature pipeline (grey_search): OpenAlex, ClinicalTrials.gov, and domain-constrained seed-site crawling (ERC/ILCOR/AHA/RCUK) using two query sets (A_high_recall; B_instruments), followed by relevance ranking and stopping rules
search_strategies
Data collection procedures
Database search strategy architecture
Query A = high-recall HRQoL strategy (cardiac arrest/resuscitation terms AND HRQoL/QoL/PROM concepts; in PubMed includes survivorship/follow-up terms as well).
search_strategies
Query B = high-precision strategy adding an instrument filter (e.g., EQ-5D, SF-36/12, RAND-36, SF-6D, HUI, 15D, WHOQOL, PROMIS, VR-12/36, QWB, SIP, NHP, Duke Health Profile, QOLIBRI).
search_strategies
Note: Web of Science uses a single combined boolean that already includes follow-up/survivorship terms.
search_strategies
Grey literature searching
Implemented via grey_search pipeline with two query sets:
A_high_recall (cardiac arrest AND HRQoL/QoL/PROM AND survivorship/follow-up/long-term terms)
B_instruments (cardiac arrest AND instrument names AND HRQoL terms)
search_strategies
Sources: OpenAlex; ClinicalTrials.gov; domain-constrained crawling for ERC, ILCOR, AHA, RCUK.
search_strategies
Ranking filters: include-term boosting; exclusion terms (animal/rat/mice/pediatric); minimum score threshold.
search_strategies
Volume controls & early stopping: max records/query 200; warm-up 100; early-stop after 50 consecutive likely-irrelevant hits; max “Google-like pages” 10 if enabled.
search_strategies
Full search strategies provided here: https://github.com/Brickfielder/QoL-cardiac-arrest/blob/main/search_strategies.md
Deduplication
Deduplicate across all sources (databases + grey outputs) prior to screening, using a reference manager and/or scripted workflow, and document the method used.
Screening workflow
Title/abstract screening.
Full-text screening (where accessible) or “abstract-only inclusion” flagged where full text unobtainable but eligibility is clear.
Reviewer procedures / QA
Calibration set at start (e.g., 100–200 records) to align eligibility interpretation.
Dual screening of a subset (e.g., 10–20%) and all “uncertain” records with consensus resolution; disagreements adjudicated by a third reviewer where needed.
Sample size
Not fixed; equals number of eligible studies identified.
Sample size rationale
Scoping review — goal is mapping breadth and heterogeneity, not powering an effect estimate.
Stopping rule
Databases: stop after all planned queries (A and B where applicable) have been executed and results exported, plus any planned citation chasing.
Grey literature: stop according to the pipeline’s predefined stopping/volume rules (max records/query; warm-up threshold; early-stop on consecutive irrelevance), plus minimum-score filtering.
search_strategies
Variables
Manipulated variables
None.
Measured variables (data charting fields)
Study identification
Citation, year, country/region, publication type (journal article, conference, report, thesis, guideline-related output), and source (PubMed/CINAHL/WoS/grey pipeline).
Study type / design
RCT, cohort, registry analysis, cross-sectional, qualitative/mixed methods, feasibility/service evaluation, etc.
Population
OHCA vs IHCA (explicit; otherwise mixed/unclear)
Adults vs paediatrics (and how handled)
Sample size; age/sex summary if reported.
HRQoL instrument(s)
PROM name and version (e.g., EQ-5D-3L vs 5L; SF-36 vs SF-12; PROMIS Global etc.)
Reporter (patient vs proxy) if stated
Mode (postal/phone/clinic/online) if stated.
Timing
Exact follow-up time(s) as reported; anchor point if clear (arrest date vs discharge vs enrolment).
Harmonised time-window bin (see Transformations).
Outcome reporting
Whether HRQoL is primary/secondary/descriptive
Whether results are presented as index scores, domain profiles, VAS, etc.
Attrition / missingness
Follow-up completion and attrition reporting where available.
Indices (derived descriptive mappings)
PROM × time-window matrix (counts)
Time-window distribution across the literature
Stratified maps by OHCA/IHCA where feasible
“Instrument diversity over time” (e.g., distinct PROMs per 5-year band)
Analysis Plan
Statistical models
Primary analysis is descriptive:
Frequencies/proportions of PROMs used overall and by subgroup (OHCA/IHCA, era, setting, study design).
Cross-tabulations of PROM by time-window.
Narrative synthesis of patterns and gaps.
No meta-analysis planned.
Transformations
Follow-up time harmonisation
Convert reported timepoints into prespecified bins, for example:
In-hospital / discharge
0–1 month
1–3 months
3–6 months
6–12 months
12–24 months
24 months
(If a study reports ranges, code using the closest matching bin and retain the original wording in a separate field.)
Inference criteria
No null-hypothesis significance testing planned for primary objectives. Any inferential comparisons (if later added) will be explicitly labelled exploratory.
Data exclusion
Exclude:
Animal-only studies.
Studies not involving cardiac arrest survivors (or mixed populations without separable data).
Studies without identifiable HRQoL PROM measurement (unless you explicitly decide to chart qualitative QoL-only separately).
Missing data
Code as “not reported/unclear”; do not impute.
If full text unavailable but abstract clearly identifies PROM and timing, include but flag as “abstract-only”.
Exploratory analysis
Trends in PROM use over time.
Differences in instrument/timing patterns by OHCA vs IHCA when classification is possible.
Relationship between follow-up intensity (number of HRQoL assessments) and attrition reporting (if consistently available).
Other
Risk of bias / quality appraisal
Not planned (typical for scoping reviews), because the aim is mapping measures and timing rather than estimating effects. If later added, it will be documented as a protocol deviation.
Overlap handling
Link multiple reports from the same cohort/registry to avoid double-counting; treat as one “study family” while capturing all unique PROMs/timepoints reported.
Data & code transparency
Share extraction template, PROM–timepoint map, and analysis code (where licensing allows) via OSF/GitHub
ADAN: Toward Structural Conditions for Machine Consciousness
No full textADAN (Autonomous Dynamically Adaptive Network) is a
continuous-time dynamical architecture that formally
implements the structural conditions theorized by leading
neuroscientific and philosophical frameworks — Integrated
Information Theory (IIT), Global Workspace Theory (GWT),
the Free Energy Principle (FEP), and enactivist autopoiesis
— as necessary prerequisites for machine consciousness.
The architecture implements seven structural conditions:
global integration via a recurrent dynamical workspace,
homeostatic viability regulation via Control Barrier
Functions (CBF) and Control Lyapunov Functions (CLF),
a tractable integration proxy derived from the Jacobian
Frobenius norm, an explicit metabolic energy budget,
viability-gated Hebbian plasticity, criticality regulation
targeting the edge-of-chaos regime, and a monotonic
entropic temporal arrow coupled to causal identity
continuity.
No claim of phenomenal consciousness is made. ADAN
constitutes a falsifiable, implementable structural
substrate from which such questions may be empirically
investigated.
Author: Bladimir García, Independent Researcher,
Dominican Republic, 2026
HOLISTIC THINKING, SHARED REALITY, AND MEANING-MAKING AMONG INDONESIAN WORKING YOUNG ADULT
No full textCerumen Tau as a Zero-Burden Biomarker for Alzheimer's Disease: A Mechanistic Hypothesis Grounded in TMC1/TMC2 Scramblase Dysfunction
No full textThis document establishes priority for two hypotheses, dated 2026-02-22.
Primary hypothesis: Cochlear hair cell death and Alzheimer's disease neurodegeneration share a common upstream cause in systemic membrane lipid dysregulation (PE/PC depletion), mediated by TMC1/TMC2 scramblase dysfunction (a mechanism identified by NIDCD researchers in February 2026). This predicts that otoacoustic emission (OAE) decline rate, plasma phospholipid ratio (PE/PC), and plasma pTau181 form a correlated biomarker cluster tracking the same pathological process, with OAE decline as the earliest accessible signal, independent of auditory deprivation.
Secondary hypothesis (priority claim): Cerumen (earwax) has not previously been investigated as a tau biomarker matrix. This document proposes cerumen tau (detected via EVs released during cochlear hair cell apoptosis and incorporated into ceruminous secretions via pericochlear vasculature) as a zero-burden, non-invasive, route to cochlear and CNS-adjacent tau burden
“Environmental Modulators of Bioenergetic Coherence: Light, Oxygen, and Circadian Integration (ISHEA-Bio)”
No full textThis component investigates the systemic effects of sunlight, outdoor exposure, oxygenation, and circadian rhythms on bioenergetic coherence. Using the ISHEA-Bio framework, it integrates evidence from human cohorts, photobiomodulation studies, and oxygenation research to demonstrate how these environmental factors influence NAD⁺/NADH balance, ROS regulation, ATP production, protein folding, and systemic regeneration.
The work differentiates Upstream factors (light exposure, circadian alignment, air quality) from Downstream outcomes (ATP availability, folding integrity, enzymatic coherence, regenerative processes). Temporary interventions like LED or red-light therapy are contrasted with sustained natural exposure effects.
This component is suitable as Upstream reference in Two-Level Doc structures for:
Bioenergetic modeling
Experimental planning
OSF preprints
ISHEA system integratio
Nanoparticle-Based Vaginal Drug Delivery Platforms for the Treatment of Candidiasis: A Scoping Review Protocol
No full textVulvovaginal candidiasis (VVC) is one of the most prevalent mucosal fungal infections worldwide, primarily caused by Candida albicans, although non-albicans species such as Candida glabrata are increasingly reported. Current treatment strategies rely mainly on topical azoles or oral Fluconazole. While these regimens achieve high cure rates in uncomplicated cases, therapeutic failures, recurrent infections, drug resistance, and limited options during pregnancy highlight important clinical challenges. In this context, vaginal drug delivery systems incorporating soft nanoparticles have emerged as promising strategies to enhance antifungal efficacy. These platforms may improve drug solubility, mucosal penetration, retention time, controlled release, and local bioavailability while potentially reducing systemic exposure and adverse effects. This scoping review aims to map and synthesize the available evidence regarding pharmaceutical platforms containing soft nanoparticles for vaginal administration in the treatment of VVC, evaluating their reported therapeutic advantages and identifying translational challenges for clinical application. The findings are expected to provide a comprehensive overview of current technological advances and research gaps in the field of nanotechnology-based vaginal antifungal therapy
Evolution and Spread of Ideology-Related Keyword Usage Across Disciplines: The Surge and Dissemination of ‘Gender’ and ‘Green’
No full textIt remains unclear how different academic disciplines conceptually interpret the social world. This study analyzed 1,283,449 abstracts published between 2002 and 2024 across twelve disciplines: Sociology, History, the Humanities, Philosophy, Linguistics, Politics, Economics, Medicine, Biology, Chemistry, Physics, and Mathematics. Through a bibliometric analysis of 26,073 Sociology abstracts, this study identified 23 “top differentially frequent ideology-related keywords” (Top DIKs). Over the study period, ‘gender’ and ‘women’ rose to prominence, whereas keywords traditionally associated with macroscopic social structures or collective solidarity, such as ‘EU’ and ‘minority,’ exhibited a decline. The frequency of these gender-related keywords in Sociology increased linearly, eventually appearing in 7–9% of all publications. Notably, ‘gender’ showed one of the highest rates of increase among all analyzed terms, transcending its specific classification as an ideology-related term. Other disciplines followed this upward trend starting in the early 2010s, albeit with lower absolute frequencies (reaching 1–4% in the human sciences and 0.2–2% in the natural and mathematical sciences). Furthermore, across all twelve disciplines, ‘women’ appeared approximately twice as frequently as ‘men,’ reflecting a persistent asymmetry in research focus. This trend was accompanied by relatively less academic attention to physical social risks compared to their real-world prevalence.
Local context-level co-occurrence analysis categorized these disciplines into five distinct clusters, revealing shared conceptual vocabularies. Group 1 (Sociology and the Humanities), Group 2 (History, Philosophy, and Linguistics), and Group 3 (Politics and Economics) collectively formed a broad human sciences cluster, while Group 4 (Biology and Medicine) and Group 5 (Physics and Mathematics) aligned as the natural and mathematical sciences. Although Chemistry shared substantial similarities with Biology, its divergence from Medicine precluded the formation of a single cohesive cluster. This classification highlights a profound affinity between Sociology and the Humanities regarding their ideology-related vocabularies. This connection is further supported by the contextual diversity of gender-related keywords: for instance, while Medicine and Chemistry emphasized women’s healthcare and reproduction, Sociology focused on women’s rights.
In contrast to these trends, the usage of ‘green’ in Chemistry showed a linear increase throughout the study period, reaching 3.5% of publications. Meanwhile, the human sciences (Sociology, the Humanities, Economics, and Politics) exhibited rapid growth after 2019, nearly converging with the levels observed in Chemistry. Co-occurrence patterns revealed a notable semantic divergence between Chemistry and the human sciences. Regarding ‘green,’ Economics and the Humanities formed a tight cluster, contrasting with the patterns for ‘gender’ and ‘women,’ where Sociology and the Humanities were more closely aligned.
Collectively, these findings offer a data-driven framework for understanding the evolution and dissemination of ideology-related keywords, underscoring the structural dominance of specific issues. These observations reflect a shift in thematic priorities, signaling a transformation in how academic disciplines engage with societal concerns. Moreover, the ubiquity of these terms masks divergent conceptual frameworks across disciplines, highlighting the necessity of contextualizing interpretations within interdisciplinary research.
DOI: https://doi.org/10.5281/zenodo.1873806