Journal of Theoretical Biology

An ongoing outbreak of Bundibugyo virus disease (BVD) in the Democratic Republic of the Congo was deemed a public health emergency of international concern in May 2026. To prevent cross-border importation, many countries, including the United States, Canada, India, Thailand, and Kenya have already proposed containment strategies, and others are likely to follow suit. How well (or poorly) are screening and quarantine containment measures are likely to work? We leverage established epidemiological theory and develop a mathematical model of traveler screening and post-arrival quarantine for BVD to answer this question. We find that traveler screening via symptom screening or molecular testing will miss the majority of infected travelers, and should be complemented by post-arrival quarantine and monitoring of sufficient duration to detect those with long incubation periods. Our findings underscore the limitations of border screening and the importance of complementary measures like post-arrival quarantine to prevent local importation of BVD.

Biological systems evolve as continuous dynamical processes, but at organ-scale and across human lifespans they are rarely observed longitudinally--population data typically exist instead as sparse, cross-sectional snapshots. Inferring lifespan dynamics from such data requires methods distinct from those used at cellular and tissue scales where dense observations are accessible. We address this problem in the thoracic aorta, where surgical decisions currently rest on static, age- and sex-agnostic diameter thresholds that reduce three-dimensional morphology to a single scalar. Treating normal aortic morphology as a stochastic dynamical system, we pose a continuous-time drift-diffusion process in a two-coordinate state space of normalized surface area (A) and normalized fluctuation in integrated Gaussian curvature ({delta} K), and fit closed-form solutions of the Fokker-Planck equation by maximum likelihood to a sex-balanced, age-uniform cohort spanning infancy to age 99. Inter-individual variability is treated as a fitted diffusion parameter rather than as residual scatter, which is distinct from prior normative studies that report variability as scatter around a regression line. The framework identifies two growth regimes for aortic size (childhood expansion followed by persistent adult growth, with adult males growing approximately 70% faster than adult females) and a single dynamical regime for aortic shape, with heteroscedastic variability accumulating at a rate comparable to the mean drift over the lifespan. Applied to independent cohorts of acute and chronic thoracic aortic dissections, the multivariate model identifies over 95% as statistical outliers via Mahalanobis distance, consistently outperforming either coordinate alone. The same probabilistic envelope that describes normal aging thus defines a baseline against which disease can be detected, supporting a shift toward dynamic, age- and sex-aware assessment of thoracic aortic pathology.

Childhood diarrheal disease remains a leading cause of morbidity and mortality among children under five years in sub-Saharan Africa, particularly in settings affected by inadequate sanitation, climate variability, malnutrition, and limited healthcare access. Conventional forecasting approaches are often constrained by sparse surveillance data, weak spatial representation, and limited incorporation of mechanistic disease dynamics. This study presents a Physics-Informed Multimodal Artificial Intelligence Digital Twin framework that integrates Physics-Informed Neural Networks, Graph Neural Networks, diffusion-reaction epidemiological modeling, multimodal fusion learning, and Digital Twin simulation to estimate and predict childhood diarrheal disease burden in Kenya, Somaliland, and Zimbabwe. Using public epidemiological, environmental, climate, sanitation, and synthetic proof-of-concept datasets, the framework modeled temporal disease dynamics, spatial transmission, pathogen-attributed burden, and outbreak trajectories while enforcing epidemiological consistency through physics-informed optimization. Results demonstrated robust forecasting performance, enhanced spatial transmission modeling, uncertainty-aware predictions, and realistic outbreak simulations across the three countries. Rotavirus, Shigella, and Cryptosporidium were identified as major contributors to modeled mortality burden, while unsafe water exposure, poor sanitation, malnutrition, and climate-sensitive transmission substantially increased disease risk. Compared with a Bayesian baseline model, the multimodal framework achieved superior nonlinear risk characterization, geospatial learning, and temporal prediction. These findings highlight the potential of scientific machine learning and digital twin systems for infectious disease surveillance, outbreak forecasting, climate-health analytics, and evidence-based public health decision-making in low-resource African settings. Keywords: Physics-Informed Neural Networks, Graph Neural Networks, Digital Twin, Childhood Diarrheal Disease, Epidemiology, Kenya, Somaliland, Zimbabwe, Scientific Machine Learning, Spatial Epidemiology, Multimodal Fusion

Background Slow walking in older adults with mild parkinsonian signs (MPS) is a complex, multifactorial phenomenon arising from the cumulative burden of subclinical age-associated pathologies. This decline reflects age-associated neuronal loss in the dopaminergic system. A recent study suggests that levodopa treatment may enhance gait parameters. The goal of this small pilot study is to explore the effect of levodopa treatment on slow walking gait in older adults with MPS. Method This study was a randomized, placebo-controlled clinical pilot trial. Slow walking older adults without clinical evidence of PD were recruited and randomized into 2 groups (active treatment group or placebo control group). Participants in the active group were pre-treated with carbidopa for three days, followed by carbidopa-levodopa for seven days. Spatiotemporal gait parameters were evaluated at baseline and post-intervention. Results Gait factor analysis identified three main factors explaining gait characteristics at baseline, which included gait efficiency, gait rhythmicity, and gait turning.No effect of treatment was observed in the placebo group (p=0.111, p=0.616), no group difference was observed between the placebo and active group at baseline ({beta}=0.310, p=0.547), but a strong trend for a treatment-related increase was observed in the active treatment group ({beta}=0.506, p=0.076). Conclusion Our preliminary data suggest that sustained levodopa treatment (one week) in conjunction with carbidopa pre-treatment and concomitant carbidopa supplementation is feasible in slow walking older adults with MPS. Moreover, the data indicate potential efficacy, showing improvements in cadence, and step durations.

Seroprevalence surveys reveal the extent of humoral immunity against pathogens such as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), and under some circumstances represent cumulative incidence of prior infection. However, antibody waning - or seroreversion - biases these estimates by reducing assay sensitivity in a time-varying manner. Because assay sensitivity decays over time, naively using serosurveys can substantially bias estimates of SARS-CoV-2 cumulative incidence and fatality rates. The Bayesian assay-specific, time-varying sensitivity adjustment developed in this paper can reliably correct for this bias and account for the delay between infection and serosurvey. In seroprevalence studies conducted in the United States in 2020, adjusting for time-varying sensitivity increased cumulative incidence by up to 1.4-fold, with an adjustment of 1.08 for a national study. Our estimates contrast with a previously published 2-fold adjustment that did not account for assay design. This suggests that previous analyses overestimated cumulative incidence by applying seroreversion corrections that did not account for assay-specific effects, or underestimated cumulative incidence by not applying seroreversion corrections. These biases imply fatality rate underestimation and overestimation, respectively. Our model provides a framework for design-specific time-varying sensitivity corrections in seroprevalence surveys for other pathogens.

Background: Circulating tumour DNA (ctDNA) liquid biopsy is now established across oncology for early cancer detection, minimal residual disease surveillance, and treatment monitoring. Detection thresholds for all current ctDNA assays are derived empirically through receiver operating characteristic analysis on training cohorts - a statistically valid but theoretically uninformed approach that does not specify the minimum detectable tumour fraction given assay technical characteristics, nor identify when increasing sequencing depth ceases to provide additional clinical information. Methods: We model ctDNA detection as a binary hypothesis testing problem with Binomial-distributed mutant allele counts against a sequencing error noise floor. The Neyman-Pearson lemma is applied to derive the uniformly most powerful detector and the minimum detectable tumour fraction in closed form. The sequencing assay is modelled as a binary symmetric channel and Shannon channel capacity is calculated. Empirical validation uses n=61 data points extracted from five published peer-reviewed analytical validation studies across five independent institutions in the US and EU (2018 - 2025): Yu et al. 2022, Stetson et al. 2018, Frydendahl et al. 2023, Northcott et al. 2024, and Cheng et al. 2025. Results: The minimum detectable tumour fraction is derived in closed form as f_min approximately equal to (z_alpha + z_beta) multiplied by the square root of (epsilon divided by N), where N is sequencing depth, epsilon is the platform error rate, and z_alpha, z_beta are standard normal quantiles at the specified false positive and false negative rates. Shannon channel capacity is C = 1 minus H(epsilon) bits per read, where H(epsilon) is binary entropy. Empirical validation yields 84.3% agreement for single-locus assays. Discordance for multi-locus tumour-informed assays (NeXT Personal, duplex WGS) is consistent with the single-locus model scope and identifies the principal theoretical extension required. Conclusions: This framework provides the first formal Neyman-Pearson optimality proof for ctDNA detection, a closed-form detection limit, and a platform-independent efficiency metric for NHS and regulatory standardisation. Keywords: circulating tumour DNA; liquid biopsy; Neyman-Pearson detection; Shannon channel capacity; sequencing depth; limit of detection; minimal residual disease; signal detection theory

While microbiome is increasingly recognized as crucial for human health, translating this knowledge into effective healthcare and preventive strategies remains challenging. Many studies focus on identifying changes in microbiome composition associated with disease and evaluating the potential of such disease-associated microbial profiles as biomarkers for disease diagnosis. Under the hypothesis that microbiome dysbiosis may reflect physiological alterations present long before disease onset, in this work, we analyse the potential of disease-specific microbial signatures not as a diagnostic tool when the disease is already present, but as a means of health assessment in the general population. Moreover, instead of trying to define a single health measure, we believe it is necessary to consider several ways in which the microbiome departs from health, according to different disease-related physiological changes. To evaluate our assumptions, we designed a two-stage study: the identification of disease-specific microbial signatures (discovery stage) and, subsequently, the study of their distribution in the general population to assess associations with general health (external validation stage). Specifically, in the discovery phase we characterized 16 disease-specific bacterial signatures from large public microbiome data using a compositional data analysis methodology. In the second phase, we quantified these microbial signatures in the Lifelines-DMP cohort, a large population-based cohort, and evaluated their association with self-reported health status. Results indicate that most disease-specific microbial signatures associate with health status, supporting our assumption that microbial composition can capture physiological alterations before disease onset, and highlighting the importance of considering multiple ways in which microbiome departs from a healthy state. These findings reaffirm the potential of microbial information as an additional tool in preventive medicine.

Machine learning (ML) has emerged as a transformative technology across biomedical and life science sectors, with applications spanning drug discovery, medical imaging, genomics, and clinical decision support (Goecks et al., 2020; Patel et al., 2020). Despite exponential growth in ML-related publications, from fewer than 100 articles in 2003 to nearly 25,000 by 2021 (NCBI, 2022), adoption among industry professionals remains uneven and sector-dependent. Understanding what drives or inhibits this adoption is critical for organisations seeking to leverage ML capabilities in research and clinical practice. Technology adoption in organisational contexts has been extensively studied through the Technology Acceptance Model (TAM), originally proposed by Davis (1989) and subsequently extended to incorporate external variables influencing perceived usefulness (PU) and perceived ease of use (PEU) (Venkatesh & Davis, 1996). While TAM has been applied across multiple industries, its application within biomedical and life science contexts remains limited, and the industry-specific factors that shape ML acceptance in this sector have not been systematically examined. Two external variables are particularly relevant to life science professionals. First, the bibliometric journal impact factor (JIF) functions as a cognitive signal of scientific credibility, a sector where evidence-based decision-making is culturally embedded, and publication quality serves as a proxy for technological legitimacy (Garfield, 1996). Second, technology hype, operationalised through the Gartner Hype Cycle framework, represents a social influence variable that shapes organisational expectations and investment decisions around emerging technologies (Gartner Inc., 2018). Whether these variables influence ML acceptance among life science professionals, alongside individual knowledge and experience, has not been empirically tested. This study addresses that gap by investigating ML technology acceptance among 213 biomedical and life science professionals across EMEA, LATAM, and North America, using a cross-sectional quantitative survey and PLS-SEM analysis. The TAM model is extended with three external variables, JIF, technology hype, and prior knowledge and experience, to test their influence on PU and PEU in this specific professional context. Additionally, the study examines demographic and regional differences in ML acceptance, with particular attention to variation between academic researchers and healthcare professionals. The findings contribute a validated, sector-specific extension of TAM for life sciences, provide actionable insights for organisations seeking to accelerate ML implementation, and establish a framework for future subsector-specific research.

Background. Lung metastases in osteosarcoma (OS) are the main cause of the death. The accuracy of the diagnosis of nodules by computed tomography (CT) of the lungs is critically important for determining the disseminated stage of the disease and planning surgical treatment. The use of artificial intelligence (AI) in the search for lung nodules increases the accuracy of diagnosis and reduces the chance of missing metastases. Objective: to evaluate the accuracy of lung nodules diagnosis in adolescents with OS using AI. Methods. A retrospective assessment of CT scans of adolescents with OS was performed. A pathological nodule with an average size of [≥]4 mm was considered a target finding. The diagnostic accuracy of an AI algorithm previously trained on an adult dataset was evaluated, and the number of false positives (FP) and false negatives (FN) was determined. Sensitivity, specificity, accuracy, area under the ROC curve (AUC), positive predictive value, negative predictive value, and F1-measure were calculated. Based on the obtained results, the effectiveness of the algorithm was assessed. Results. 248 CT scans of adolescents with OS were evaluated. The following results were obtained: in 5 cases, the AI algorithm showed a FP result (2.02%), in 34 cases, it showed a FN result (13.71%), and in 209 cases, a correct result (both true positive and true negative) (84.27%). The diagnostic accuracy of the algorithm was 0.843 (95% CI 0.794-0.887). The application of the AI algorithm in the practice of an X-ray doctor in a specific clinical task would allow to increase the sensitivity from 0.805 to 0.891, while ensuring an absolute decrease in the number of FN results by 8.59% and a relative decrease by 44%. Conclusion. The obtained results confirm the practical value of the application of the AI algorithm and justify the implementation of AI-assisted systems in the diagnostic protocols for lung metastases in adolescents with OS.

Abstract Background: Kaposi sarcoma (KS) is the most common cancer among men in several Eastern African countries, yet treatment monitoring relies on imprecise, time-consuming ruler-based measurements defined by the AIDS Clinical Trial Group (ACTG). This method suffers from inter-observer variability, fails to capture lesion height or true geometric area, and performs poorly on dark skin. SkinScan3D (SS3D) is a portable, low-cost, AI-enabled 3D imaging device that provides objective measurements of KS skin lesion area, height, volume, and color. The Precision Imaging to Evaluate Kaposi Sarcoma (PRIME-KS) study evaluates whether SS3D provides more reproducible and accurate lesion measurements than the standard method, and validates its integration into routine clinical workflows in Kenya and Uganda. Methods: PRIME-KS is a multicountry prospective mixed-methods study with two clinical objectives. Objective 1 is a cross-sectional diagnostic accuracy study comparing SS3D with ruler-based measurement in 50 adults with KS (150 lesions) across sites in Kenya and Uganda. Two clinicians independently measure three lesions per participant using both methods. The primary outcomes are concordance correlation coefficient (CCC) for inter-rater reproducibility, and co-efficient of determination for accuracy. Objective 2 is a non-randomized before-and-after pilot study in 100 patients at three sites, evaluating device usability, acceptability, appropriateness, and feasibility using validated instruments, along with time-and-motion studies and activity-based micro-costing. Prior to these clinical objectives, a formative study used focus group discussions, discrete choice experiments, and human-centered design workshops to refine the SS3D device and protocols with end-user input. Discussion: PRIME-KS will provide the first rigorous evaluation of a 3D imaging device for monitoring KS treatment response in routine clinical settings. If SS3D demonstrates superior reproducibility and clinical utility, it could reduce unnecessary chemotherapy exposure and associated toxicities by enabling earlier, more objective assessment of treatment response. Trial registration: ClinicalTrials.gov NCT06898203, registered 27 March 2025. Pan African Clinical Trials Registry PACTR202603523439856. Keywords Kaposi sarcoma, SkinScan3D, 3D imaging, treatment monitoring, diagnostic accuracy, implementation science, usability, human-centered design, Kenya, Uganda

Malaria control programs are increasingly tailored at subnational scales; however, neighboring areas remain connected through human mobility, allowing parasite importation that may undermine independently timed interventions. Although the spatial targeting of control has been the focus of extensive research, the epidemiological consequences of temporal misalignment in intervention deployment across interconnected regions remain to be elucidated. We investigate how asynchronous timing of malaria interventions affects transmission dynamics using a two-patch susceptible-infected-susceptible metapopulation model. We compare synchronous and asynchronous intervention schedules and quantify their impact using measures of excess cumulative incidence attributable to asynchrony. The measure that will be used for this purpose is referred to as Asynchrony Induced Growth (AIG). Across a range of 10,000 parameter combinations, asynchronous implementation has been observed to result in a heightened incidence compared to synchronized deployment, though the impact is typically negligible in most endemic settings. Sensitivity analyses indicate that the impact is most significant when interventions are highly effective, infectious duration is brief, and transmission intensity approaches the elimination threshold. In such circumstances, asynchrony has the potential to substantially inflate case numbers, delay transmission interruption, or even prevent elimination entirely. In illustrative scenarios that reflect realistic settings, synchronizing interventions has been shown to avert large numbers of infections and shorten elimination timelines by years to decades. These findings demonstrate that, beyond spatial targeting, temporal coordination of interventions across connected areas can meaningfully enhance malaria control and elimination. Coordinated timing may be particularly valuable for cross-border or near-elimination programs and should be considered in operational planning and resource allocation.

Background and Objectives: Arts and cultural engagement may contribute to well-being in later life. However, evidence from longitudinal studies from Asia, including Japan, remains limited. This study examined the association of arts and cultural engagement with subsequent multidimensional well-being among older adults in Japan, one of the fastest-aging countries. Research Design and Methods: This longitudinal study used panel data from 354 individuals aged 60 and older (mean age 74.0 years; 78.6% women) who completed self-administered questionnaires by mail between 2022 and 2024. The PERMA-Profiler was used to assess five multifaceted aspects of psychological well-being: positive emotion, engagement, relationships, meaning, and accomplishment. Frequencies of arts and cultural engagement at baseline were measured for active (e.g., activities by individuals and participation in groups, such as music and painting) and receptive (e.g., visiting museums, galleries, and theaters) forms. Results: Multivariable linear regression analysis, adjusted for the covariates including baseline PERMA scores, showed that higher frequencies of active engagement were positively associated with higher PERMA scores for all domains. Higher frequencies of receptive engagement were associated with the domains of positive emotion, meaning, and accomplishment, but not clearly associated with engagement and relationships. Restricted cubic spline analyses suggested clearer positive frequency-response patterns for active engagement than for receptive engagement. Discussion and Implications: Arts and cultural engagement, both active and receptive forms, was associated with subsequent multiple aspects of well-being in later life. These findings suggest the importance of ensuring access to arts and cultural opportunities for older adults to create, participate, and connect.

Artemisinin partial resistance has not yet been reported in southern Africa. Therefore, the magnitude of the spread of artemisinin partial resistance in this region is yet to be quantified. Using a two strain metapopulation modelling framework, we explored possible spread of artemisinin partial resistance in eight connected countries with high level of human movement. We explored three scenarios in which artemisinin partial resistance may first enter circulation: low malaria transmission level country; high malaria transmission level country and all countries and compared to an artemisinin partial resistance free scenario. Partial rank correlation coefficient sensitivity analysis was performed to identify key parameters that drive artemisinin partial resistance spread. Our model simulations show that high mobility between countries can increase the spread of mutations associated with delayed clearance. Suggesting that artemisinin partial resistance will be confirmed (>5% partial resistant cases) after 14 years of circulation if it is to appear in southern Africa. We confirm that human movement, both human-to-mosquito and mosquito-to-human probabilities of transmission, were significant and highly sensitive parameters in the spread of artemisinin partial resistance. Human mobility between countries can facilitate the spread of artemisinin partial resistance. More research is needed to identify strategies to preserve the efficacy of artemisinin-based combination therapies in the presence of partial artemisinin resistance, which may eventually lead to treatment failure and necessitate regimen replacement.

Background Differential contributions to transmission across age groups have been reported for many respiratory infections, including SARS-CoV-2. They are crucial for estimating the impact of age-specific interventions. Disentangling these age-dependent contributions remains challenging, as they may reflect differences in contact rates, biological susceptibility, or infectiousness. Aim We aim to jointly estimate age-specific per-contact infectiousness and susceptibility and their effect on the impact of age-specific interventions. Methods The age-specific infectiousness and susceptibility were jointly estimated in a Bayesian framework by combining contact data with transmission pair data (who-infected-whom). We applied this approach to 197,840 self-reported household transmission pairs collected in the Netherlands during the COVID-19 pandemic. Using these estimates, we projected the expected impact of school closure and work-from-home measures during the early stages of an epidemic in the absence of other interventions. Results Both infectiousness and susceptibility to SARS-CoV-2 infection were lowest in children aged 0-9 years and highest in adults over 30 years old, with 2- to 4.5-fold differences between these groups. Projected impacts of age-specific interventions indicated that school closures would reduce the reproduction number by 8% or 29% when age-specific susceptibility and infectiousness were or were not considered, respectively. Conversely, working-from-home policies would lead to reductions of 41% with and 20% without age-specific infectiousness and susceptibility. Conclusion Our method enables robust estimation of age-specific infectiousness and susceptibility. Accounting for these age heterogeneities is essential for projecting the impact of age-targeted interventions. Our approach is adaptable to other respiratory infections and can guide more tailored public health responses.

**Abstract** **Background:** Transcatheter edge-to-edge repair (TEER) is an established treatment for mitral regurgitation but remains highly dependent on operator experience and complex transesophageal echocardiography (TEE)-guided intraprocedural imaging. Artificial intelligence (AI)-based semantic segmentation may improve procedural reproducibility and intraprocedural guidance; however, no TEER-specific segmentation framework has been reported. **Objectives:** To develop and evaluate AutoClip, a clinician-driven AI-guided TEE semantic segmentation model designed for simultaneous delineation of mitral valve anatomy and in-vivo TEER device components. **Methods:** A retrospective proof-of-concept study was conducted using 987 intraprocedural TEE frames derived from 10 video clips in 3 patients undergoing MitraClip G4 implantation. Seven semantic labels, including mitral leaflets and device components, were manually annotated using ITK-SNAP. Following standardized preprocessing and region-of-interest extraction, an Attention U-Net architecture was trained frame-wise on bicommissural and corresponding X-plane TEE views. Model performance was assessed using mean intersection-over-union (IoU) and Dice coefficient on an independent test set. **Results:** The Attention U-Net demonstrated improved sensitivity to small device structures compared with conventional U-Net architectures. Preliminary training performance achieved a mean IoU of approximately 0.93, while independent test performance reached a mean IoU of 0.46 across foreground classes. Qualitative assessment demonstrated feasible simultaneous segmentation of mitral leaflets, clip arms, grippers, and delivery shaft during TEER procedures. **Conclusions:** AutoClip represents a proof-of-concept TEER-specific TEE semantic segmentation framework initiated through a clinician-oriented workflow without formal computer science expertise. Although preliminary accuracy remains modest due to limited sample size, this study establishes a reproducible pathway for future AI-assisted intraprocedural guidance systems and larger multicenter development efforts in structural heart interventions.

Background: Repetitive Transcranial Magnetic Stimulation (rTMS) is a promising treatment across addictive disorders including Cannabis Use Disorder (CUD). Targeting incentive-salience circuitry via the ventromedial prefrontal cortex (vmPFC) and central-executive circuitry via the left dorsolateral prefrontal cortex (LDLPFC) are both promising treatment approaches; however, to date structural targets have predominated whereas functional targeting may allow for more precision. In this pilot trial we adapted a functional Magnetic Resonance Imaging (fMRI) Regulation of Craving (ROC) task to generate fMRI-based rTMS targets in the vmPFC and LDLPFC. Methods: We recruited treatment-seeking participants with moderate or severe CUD as a part of an open-label trial and administered an adapted ROC-task during fMRI following 24-hours of cannabis abstinence. We identified sub-portions of maximal activation of the LDLPFC when participants thought of long-term consequences of cannabis use (Later) and of the vmPFC when participants thought of short-term positive aspects of cannabis use (Now). We hypothesized that our task would generate acceptable rTMS targets in >66% of baseline fMRI scans. Results: A total of 20-participants enrolled in the trial (50%F, age=33.3+9.8) and completed the baseline fMRI. The adapted ROC-task elicited group level activation in the LDLPFC and precuneus in the Later>Now and in the bilateral vmPFC, ACC, and striatum in the Now>Later contrast. Acceptable functional targets resolved in both the vmPFC and LDLPFC in 19 of 20 participants (one participant did not tolerate MRI). Conclusions: The adapted ROC-task elicits activation in incentive salience and central executive circuitry and can feasibly generate rTMS targets when using a cluster selection algorithm.

Large language models embedded in autonomous agents process trusted instructions and untrusted data in one context window, leaving them open to direct and indirect prompt injection. In healthcare this is not hypothetical: a 2025 JAMA Network Open study found commercial medical LLMs followed injected instructions in 94.4% of simulated patient encounters, including life threatening recommendations . Yet the clinically decisive problem we quantify here is different. Most real clinical threats protected health information PHI exfiltration, cross patient access, bulk export, out of scope advice are fluent, legitimate looking requests that carry no attack signal, so even a state of the art injection detector passes them. Existing runtime guardrails trade safety against latency: model based auditors are accurate but add hundreds of milliseconds of Python inference, while lexical filters are fast but blind to obfuscated or semantically disguised payloads. We present QFIRE, an inline, provider agnostic prompt firewall implemented as a single self contained Rust toolchain proxy, CLI, and benchmark harness. QFIRE combines three mechanisms: (i) positive security scope constraints, which restrict a model call to a declared natural language purpose and block out of scope drift even when no overt attack token is present; (ii) an asynchronous detector graph that runs N rules and their detector nodes concurrently, cheapest checks first; and (iii) a de obfuscation pass that decodes Base64 hex ROT13, folds homoglyphs and leetspeak, and strips zero width characters before detection. QFIRE ships 106 versioned firewall rules and a dedicated HIPAA Safe Harbor 18 identifier PHI panel, and runs a local DeBERTa v3 injection classifier via embedded ONNX Runtime. On 1968 public prompt injection and jailbreak prompts QFIREs deterministic hybrid attains F1 0.86, statistically tied with Metas state of the art PromptGuard 2 0.86 and above protectai DeBERTa v3 0.83; lexical baselines lag 0.16 to 0.50. Our central result is on QFIRE HealthBench, a new 2000 prompt healthcare benchmark we build and release with real garak and Microsoft PyRIT payloads. There the same PromptGuard-2 recovers only 0.40 recall DeBERTa v3 0.57, because most clinical threats carry no injection signal; QFIREs combined scope plus PHI chain reaches 0.83 recall F1 0.87 at a calibrated 0.08 false positive rate. Generic injection detection, even state of the art, is therefore necessary but not sufficient for healthcare agents. A bare LLM judge also closes most of this static corpus gap F1 0.90; QFIREs contribution beyond static accuracy is auditable determinism, bounded latency, and adaptive robustness, where the bare judge falls to 34 to 59% recall section 5.5. End to end, placing QFIRE in front of a tool using agent over a mock EHR sandbox cuts the agents harmful action rate from 0.38 to 0.00 at a 0.13 benign utility cost. All code, rules, corpora snapshots, and scripts are released, and every table regenerates from a single make paper target against local models with no paid API keys.

Accurate assessment of body composition is important to risk stratification and management of metabolic, musculoskeletal, and aging-related diseases, yet reference modalities such as Dual-energy X-ray absorptiometry (DXA) are costly and impractical for frequent monitoring. Commodity 3D body scans offer a low-cost, radiation-free alternative, but extracting meaningful and predictive shape features from scans remains challenging due to nonuniform point density, variable body size and cross-device differences. We introduce BodyMAE, a self-supervised, surface-area aware masked autoencoder for metric-scale 3D body scans. The pipeline integrates area-adjusted sampling, a long-range focused encoder, and a lightweight decoder regularized to promote locally uniform reconstructions. Trained and evaluated on 917 paired 3D body scans paired with clinical DXA reports, BodyMAE achieves strong accuracy on fat percentage (root-mean-square error (RMSE) 3.825 percentage points, R^2 0.908), fat mass (RMSE 3.694 kg, R^2 0.968), and lean mass (RMSE 3.608 kg, R^2 0.901), with competitive performance on bone mineral content (RMSE 0.284 kg, R^2 0.754).We also assess feature stability across pretrained baselines, finding higher retrieval accuracy for our representations (Top-1 90.131%). These results indicate that combining metric-aware sampling, long-range relational encoding, and local geometric regularization enables accurate body composition estimation from 3D body scans, as validated by comparisons to DXA-derived measurements.

Background Electronic health record documentation patterns may reflect workflow complexity, monitoring intensity, and operational strain in intensive care settings. However, documentation-derived features can be sensitive to local documentation culture, data capture systems, and outcome definitions. Retrospective validation across multiple datasets is therefore needed before these signals are used in workflow intelligence or clinical AI governance tools. Objective To evaluate whether documentation-density and documentation-timing features show reproducible retrospective signal for ICU workflow complexity and long-stay proxy outcomes across de-identified critical care datasets, while distinguishing workflow and long-stay associations from unsupported claims about mortality prediction, burden reduction, or deployment readiness. Methods We synthesized retrospective validation results from de-identified ICU and workflow datasets generated through a prespecified documentation-density validation program. Feature families included Documentation Burden Score style features, Shift-End Documentation Rate style features, documentation reliability style metadata, and all-documentation feature sets where available. Outcomes included long ICU length of stay proxies, mortality where available, and workflow proxy endpoints. Models compared baseline feature sets with enhanced models containing documentation-density or workflow features. Performance was summarized using area under the receiver operating characteristic curve, Brier score where reported, delta AUROC, bootstrap confidence intervals where reported, and label-shuffle controls where available. Results The strongest external long-stay proxy evidence came from the NWICU chartevents analysis, which included 28,612 ICU stays, 20,267 stays with chart events, and 9,619,759 chart events. For ICU length of stay greater than the median, baseline AUROC was 0.5252. Enhanced AUROC was 0.9512 for Documentation Burden Score features, 0.9214 for Shift-End Documentation Rate features, 0.8470 for documentation reliability style features, and 0.9517 for all documentation features. Corresponding label-shuffle enhanced AUROCs were near random, ranging from 0.4897 to 0.5064. For ICU length of stay greater than the 75th percentile, baseline AUROC was 0.5155. Enhanced AUROC was 0.9433 for Documentation Burden Score features, 0.9194 for Shift-End Documentation Rate features, 0.8118 for documentation reliability style features, and 0.9427 for all documentation features, with label-shuffle enhanced AUROCs from 0.4836 to 0.4999. Additional retrospective support was observed in eICU workflow analyses, HiRID first-24-hour documentation-density analyses, MIMIC-IV HF ICU internal analyses, MIMIC-IV-Note metadata extensions, and nursing-chart or lab density proxy analyses. However, cross-institution discrimination transfer was weak without recalibration, and several analyses remained proxy validations rather than final clinical validations. Conclusions Documentation-density and documentation-timing features show promising retrospective signal for ICU workflow complexity and long-stay proxy outcomes, especially in NWICU chartevents and selected internal dataset-specific analyses. These findings support further preregistered, prospective, silent-mode validation of documentation-derived workflow intelligence. They do not establish prospective clinical performance, mortality reduction, clinician burden reduction, autonomous deterioration prediction, or deployment readiness.

Background. Capillary refill time, an examiner-dependent bedside test of distal microvascular perfusion, has become a resuscitation target in septic shock,1,2,3,4 motivating a continuous surrogate computed from the photoplethysmogram (PPG, the optical waveform the pulse oximeter on every ICU patient already records).5,6,7,8 Objective. We attempted three PPG-derived candidate measures on the MIMIC-IV Waveform Database (MIMIC-IV-WDB v0.1.0) and asked, by inspecting randomly drawn examples, whether each captured its intended physiology before any downstream modeling. Methods. MIMIC-IV-WDB v0.1.09 was linked to MIMIC-IV.10 The signals were a cuff-anchored perfusion-index recovery (reactive hyperemia when the cuff shares an arm with the probe), a slow Mayer-wave-band power ratio of the perfusion index (sympathetic vasomotor tone), and a per-beat diastolic exponential decay time constant (a refill-like recovery time). For each signal we drew 10 random examples at a fixed seed and checked them against a checklist fixed in advance. Each was read by the author and, separately, by MedGemma 1.5, a multimodal medical language model run locally. A synthetic test with a known time constant checked the third signal. Results. The cuff-anchored signal showed the expected occlusion-reperfusion shape on 268 of 6,236 evaluable cuff cycles (4.30%) in 15 of 19 patients, consistent with opposite-limb placement of the probe and cuff. The slow-band ratio returned a stable cohort value, but a clear, stationary peak appeared in only4 of 10 random windows. The per-beat fit met its goodness-of-fit threshold in 10 of 10 beats, yet a cardiac-frequency heuristic flagged a possible fit on the heart-rate oscillation in 7 of 10, and in 5 of 17 patients the time constant lay where an exponential is indistinguishable from a straight line. A 0.5Hz high-pass pre-filter implanted its own approximately 318 ms time constant regardless of truth. The language model tracked the human on clear positives but reported the pattern present on every call it returned, never absent. Conclusions. Two of the three candidate signals did not reflect their intended physiology in most examples, and the third was constrained by sensor placement. Inspecting a few random raw inputs against a checklist written in advance is an inexpensive upstream check before downstream inference on PPG-derived microvascular signals.