Biology Methods and Protocols

Background. In people with type 2 diabetes (T2D), postprandial glucose weighs on glycaemic control, and carbohydrate counting is difficult. Patients can describe their meal but rarely quantify it. Objective. To assess (1) whether a small open large language model (LLM), fine-tuned on public data, estimates the carbohydrate content of a meal described in words, and (2) what this carbohydrate adds to the prediction of the postprandial glucose rise in two public cohorts. Methods. Step 1: the Qwen3-4B LLM was fine-tuned with QLoRA on Nutrition5k dishes and compared with an automatically matched USDA food composition table and with Qwen3-4B and Qwen3-8B used without training, on test dishes (internal control) and on 100 real meals of ShanghaiT2DM patients, whose reference carbohydrate was computed from the recorded quantities. Step 2: the maximum glucose rise between 0 and 120 minutes ({Delta}Gmax) was predicted by personalised gradient boosting (M2), compared with a spline linear regression and with the Chronos-2 foundation model, using only information available at meal start and person-level cross-validation, in ShanghaiT2DM (people with T2D recruited in two Shanghai hospitals; carbohydrate estimated by the LLM) and CGMacros (free-living adults in Santa Barbara, California; carbohydrate entered in the dataset, no text). Step 3: the model was transferred between datasets. 95% confidence intervals (CI) were obtained by cluster BCa bootstrap (2,000 resamples). Results. On the 100 real meals, the carbohydrate mean absolute error (MAE) of the fine-tuned model was 12.2 [9.9, 15.2] g, almost half that of the table (22.6 g; paired difference -10.5 [-16.8, -6.0] g) and nearly a third of that of the same model without fine-tuning (34.3 g), with a bias of -10.3 g; on test dishes it was 2.2 [2.0, 2.5] g. Personalised gradient boosting (M2) reached a correlation between predicted and observed rises of r = 0.63 [0.55, 0.69] in CGMacros (1,286 meals, 45 adults) and 0.49 [0.42, 0.54] in ShanghaiT2DM (2,339 meals, 99 people with T2D), about twice the linear reference (r = 0.32 and 0.23); for a rise above 50 mg/dL, its area under the ROC curve (AUROC) was 0.80 and 0.74, versus 0.64 and 0.68 with pre-meal glucose alone. Adding carbohydrate to pre-meal glucose changed r by 0.095 [0.068, 0.128] in CGMacros and by 0.004 [-0.010, 0.017] in ShanghaiT2DM; difference between datasets: 0.091 [0.060, 0.126]. Applied from one dataset to the other, the model lost correlation ({Delta}r = -0.11 [-0.17, -0.05] towards ShanghaiT2DM and -0.12 [-0.19, -0.04] towards CGMacros). The full "text [->] LLM [->] glucose" chain was evaluated in ShanghaiT2DM only. Conclusion. An open 4-billion-parameter LLM, fine-tuned with QLoRA on public data and usable offline, almost halves the error of carbohydrate counting for real meals described in words compared with an automatically matched table, with an underestimation of large meals. Personalised gradient boosting, based on pre-meal glucose and the person's usual response, doubles the correlation of a linear reference for predicting the glucose rise in both cohorts. Carbohydrate improved this prediction in CGMacros but not in ShanghaiT2DM; the two datasets differ in many respects, and the cause of this gap is not established.

INTRODUCTION: The digital clock drawing test (dCDT) is a cognitive screening tool capturing high-resolution pen-movement data. Existing approaches for classifying cognitive impairment from dCDTs rely on selected features or image-based methods. We propose representing dCDTs as structural networks to classify cognitive impairment and characterize spatial organization in clock drawings. METHODS: Weighted networks were constructed from 3,362 Framingham Heart Study participants, with and without cognitive impairment. Network representations were generated separately for command and copy conditions, with nodes representing pre-defined clock components (digits, hands, clock face) and edges capturing pairwise spatial relationships between components. Random forest models and graph convolutional networks were trained to predict cognitive impairment from the networks. Models were benchmarked against demographics-only and established summary feature baselines. RESULTS: Network-based models achieved AUC values comparable to the established summary feature baseline across command-only and copy-only conditions. Differences in spatial organization between the cognitively intact and impaired groups were greatest in the upper-right clock region, involving digits 1 and 2, the clock hands, and the surrounding clock face. Differences between the command and copy networks were larger among cognitively impaired participants. DISCUSSION: The proposed network representations provide an objective, scalable, and flexible framework for representing clock drawings for use in classification models. dCDT networks matched the predictive performance of established summary feature-based approaches and identified clock regions where spatial organization differed by cognitive status. After external validation, this approach has the potential to be incorporated within a clinical workflow to support automated and interpretable cognitive screening.

Objective: To determine whether plasma neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP) are elevated in hospitalised patients with delirium relative to patients with dementia and primary psychiatric disorders (PPD). Methods and analysis: This was a cross-sectional observational study conducted at the Royal Melbourne Hospital, Melbourne, Australia, between April and October 2024, with comparator samples from a research biobank. In Phase 1 (Delirium Discovery Cohort), we measured plasma NfL and GFAP using the Single molecule array (Simoa) platform in hospitalised patients with delirium and compared them with dementia and PPD groups. Biomarker concentrations were the primary outcome and were compared using general linear models adjusted for age, sex, and estimated glomerular filtration rate, with bootstrapped confidence intervals. Discriminative performance was assessed as a secondary outcome using receiver operating characteristic (ROC) curve analysis. In Phase 2, we applied this threshold to independent dementia and PPD groups. Results: In Phase 1 (n =176), inpatients with delirium showed higher concentrations of both NfL and GFAP than patients with dementia ({beta} = 0.78; 95% CI [0.37, 1.20] and {beta} = 0.39 [0.07, 0.73] respectively) or PPD groups ({beta} = 1.17; [0.69, 1.66] and {beta} = 0.83 [0.51, 1.24] respectively). ROC curve analysis for differentiating delirium from dementia and PPD groups yielded an optimal NfL threshold of 24.1 pg/mL (AUC = 0.89, 96% sensitivity, 70% specificity). In comparison, GFAP showed lower accuracy (AUC = 0.85), and combining NfL with GFAP (AUC = 0.90) did not significantly improve performance. In Phase 2 (n = 154), the NfL threshold of > 24 pg/ml maintained strong performance (AUC = 0.95, 96% sensitivity, and 77% specificity). Conclusions: Plasma NfL shows promise as a biomarker for delirium in hospitalised patients, distinguishing it with high sensitivity from patients with dementia and PPD. Larger multicentre studies are needed to refine cut-offs, evaluate temporal dynamics, and determine integration into acute hospital delirium pathways.

Objective A diversity of approaches and cohorts have been published in research literature for suicide risk estimation, leading to some uncertainty around which factors and conditions are most strongly contributing to risk. Methods exist for weighting estimates by propensity score, removing simplifying the interpretation of model effects. Few studies have explored the use of these tools to assess when the effect of certain factors is uncertain or inconsistent. Method A mix of 1,001,737 suicide decedents and living patients were evaluated for suicide risk, relying on a subset of commonly identified factors: demographics, psychiatric conditions, substance use disorders, and social needs. Care utilization patterns were then assessed controlling for these factors in a hospital care setting. Four methods were considered for controlling multivariate confounding of odds predicting death by suicide: multivariate logistic regression, 1-to-1 sample matching, inverse propensity weighting (IPTW), and entropy balancing (EBAL). Results Test statistics showed suicide decedents were much more likely to have various psychiatric and substance use diagnoses, social needs and more likely to heightened care utilization across sampling intervals. Greater utilization was found close to date of death. Effect sizes after IPTW and EBAL weighting were also generally greater than would have been documented by regression alone. Conclusions Propensity weighting is a useful alternative to exploring nuanced effects, especially when the direction and strength of an association differ from traditional statistics. By weighting, the overall imbalance of cases remains consistent and effect terms might reflect more nuanced associations than regression or sample matching alone.

Glucagon-like peptide-1 (GLP-1) receptor agonists produce less weight loss in people with type 2 diabetes than without, and higher compensatory doses are limited by gastrointestinal adverse events and lean-mass loss--leaving patients with obesity, type 2 diabetes, and metabolic dysfunction-associated steatohepatitis (MASH) inadequately treated. We tested whether linafexor, a pulsatile farnesoid X receptor (FXR) agonist, could augment the weight-loss efficacy of low-dose semaglutide in this population while preserving tolerability. In this multicenter, randomized, double-blind, placebo-controlled, phase 2 trial, 62 adults with obesity (body-mass index, 30-45 kg/m2), treatment-naive type 2 diabetes, and MASH (MRI-PDFF [≥]10%) were assigned to linafexor (4 mg orally once daily) or placebo, each with semaglutide (0.5 mg once weekly), for 16 weeks. The primary outcome was the percentage change in body weight at week 16. Adding linafexor more than doubled weight loss (-6.17% vs. -2.91%; treatment difference, -3.26%; 95% CI, -5.35 to -1.18; P=0.003), and more participants achieved at least 5% weight loss (53.3% vs. 28.1%; P=0.046). Consistent with this, combination therapy further reduced visceral adipose tissue (treatment difference, -13.85 cm2; P=0.035), the liver enzymes alanine aminotransferase and {gamma}-glutamyl transferase, and serum bile acids, whose changes correlated with weight and liver-enzyme improvements--associations absent with monotherapy. Gastrointestinal adverse events were comparable (23.3% vs. 18.8%), with no treatment-related serious adverse events or discontinuations. Thus, in a population with blunted GLP-1 responses, adding linafexor to low-dose semaglutide enhanced weight loss and improved body composition and hepatic measures while preserving tolerability--a dose-sparing, mechanism-complementary strategy (ClinicalTrials.gov number, NCT06492330).

Deep learning (DL) applied to haematoxylin and eosin (H&E)-stained pathology images can predict microsatellite instability (MSI) in gastric cancer (GC). We evaluated its diagnostic accuracy and discriminative performance. PubMed, Embase, Cochrane Library and Web of Science were searched for studies applying DL to H&E images for MSI prediction in GC. The protocol was prospectively registered in PROSPERO (CRD420261503140). Risk of bias was assessed with QUADAS-2. Pooled sensitivity and specificity were estimated with a bivariate random-effects model, with subgroup analyses by validation strategy and data source; slide-level areas under the curve (AUCs) were pooled after logit transformation. Prespecified scenario analyses assessed robustness. Ten studies were included; seven independent datasets from six studies entered the primary synthesis. Pooled sensitivity, specificity and summary receiver operating characteristic AUC were 0.78 (95% CI 0.63-0.88), 0.85 (95% CI 0.77-0.90) and 0.89 (95% CI 0.86-0.91); nine studies contributed to the secondary AUC synthesis (pooled AUC 0.85, 95% CI 0.80-0.89). Hospital-derived cohorts showed numerically higher sensitivity (0.86) and diagnostic odds ratio (39.8) than TCGA-derived cohorts (0.71 and 11.5), but subgroup differences were limited by the small number of studies and substantial heterogeneity. DL models show good diagnostic accuracy for MSI prediction in GC, but clinical utility still requires confirmation in standardised multicentre, independent external and prospective studies.

Aims: To understand how users believe continuous glucose monitoring (CGM) changes their behaviour, which app features they credit, and whether endorsing more mechanisms of action (MoAs) is linked to reported diet and physical activity change. Methods: We surveyed 112 U.S. adults with type 1 or type 2 diabetes who had used a Dexcom or Abbott Libre CGM within 12 months. Participants rated 26 MoA statements mapped to the Capability, Opportunity, Motivation-Behaviour (COM-B) model, identified contributing app features, and reported CGM-attributed diet and physical activity change. Logistic regression examined MoA count and COM-B domains as predictors of behaviour change. Results: Participants (mean age 54.5 years; 60.7% female) reported dietary (82.1%) and physical activity (46.4%) change and endorsed a median of 18 MoAs, most often relating to knowledge, skills and behavioural consequences. Real-time glucose and graphs were attributed to many MoAs. Each additional MoA was associated with higher odds of dietary (OR 1.27, 95% CI 1.15-1.44) and physical activity change (OR 1.27, 95% CI 1.16-1.43). Motivation made the strongest distinct contribution among COM-B domains. Conclusions: Users perceive CGM as engaging many behavioural mechanisms; endorsing more was linked to behaviour change. Motivation may be a key target for CGM-based behavioural support.

Objectives: To evaluate change in physical function and probable sarcopenia following a 36-week community multicomponent exercise programme (Vivifrail model) across two consecutive cycles, and to assess the consistency of this change between cohorts. Study design: Quasi-experimental, single-arm, repeated-measures service evaluation. Ninety-two older adults (2025-2026 cohort; mean age 75.9 years, 88.0% women) completed baseline and post-intervention assessments; an independent 2024-2025 cohort (n = 86) was analysed jointly for replication (n = 178). Main outcome measures: Short Physical Performance Battery (SPPB) score (primary); Timed Up and Go (TUG), handgrip strength, 6-metre walk time and Mini-Mental State Examination (MMSE) (secondary); probable sarcopenia (EWGSOP2), SPPB-based frailty category, and a fall-risk indicator computed identically in both cohorts. Results: SPPB improved by a mean of 1.33 points (Cohen's d = 0.96; p < .001), with concordant improvement in all four secondary outcomes (all q < .005). 72.8% reached the substantial minimal clinically important difference. Probable sarcopenia fell from 17.4% to 5.4% (p = .006), and frailty category shifted toward higher functional levels (p < .001). Effect sizes were consistent between cohorts for SPPB, handgrip and 6-metre walk; the Time x Cohort interaction was significant for TUG only. In the pooled sample, elevated fall risk fell from 19.7% to 11.2% (p = .002). Conclusions: A 36-week multicomponent exercise programme was associated with clinically meaningful, reproducible improvement in physical function and a reduction in probable sarcopenia across two independent cohorts. The single-arm design precludes causal inference.

Awareness of comorbidities in people with dementia is vital context for dementia research and clinical management. Natural Language Processing (NLP) applied to free text in the electronic health record (EHR) may help ascertain comorbidity presence. Our aim was to assess the validity of the MedCAT Comorbidity NLP app in a cohort of people with dementia. We used text from the EHR of South London and Maudsley NHS Mental Health Trust via Clinical Records Interactive Search (CRIS), linked to structured data from local primary care (Lambeth DataNet) and hospital admission data in a cohort of 3,543 people with dementia. We evaluated MedCAT Comorbidity, examining internal validity via manual review of text and external validity by cross-validation against linked primary care and hospital data; and comparison of the predictive value of comorbidity measures for mortality. Internal precision was highest for atrial fibrillation (100%) and lowest for diabetes (78%). Cross-validation with linked data varied substantially across conditions, with precision from 96% to 59% and recall 77% to 4%. The median number of comorbidities identified was two using any single source, or three when combined. MedCAT Comorbidity demonstrated predictive ability for mortality (HR 1.11 per condition), although the model fit was better using comorbidity from hospital admission data. In people with dementia, the MedCAT Comorbidity app showed high precision for most conditions during internal validation. The variability of precision and recall on external validity assessment partly reflects incomplete documentation. Results from the MedCAT comorbidity app can assist researchers in understanding comorbidity of cohorts, particularly when primary care or hospital data is not available. Our findings highlight the need to develop quality assurance protocols that include internal and external validity checks before deploying general data extraction apps for specific clinical subpopulations.

Patients with implantable cardioverter-defibrillators (ICDs) often are excluded from receiving CMR scans at 3.0T due to ICD-induced image artifacts. This work aims to transfer sequence protocols established for CMR imaging in the presence of ICDs at 1.5T to a first human application at 3.0T. All measurements were performed on a clinical 3T MRI scanner. Phantom measurements were used to establish and optimize wideband PSIR LGE imaging at 3.0T. Volunteer measurements with externally placed ICDs were used for optimization of CINE imaging and wideband PSIR imaging. Thirdly, patients with implanted ICD were examined with the optimized protocols. Human measurements showed that SA stacks obtained with spoiled GRE after contrast agent showed a markedly lower artifact burden than the conventional approach. For scar imaging, a wideband PSIR with a bandwidth of 3kHz and a frequency offset of +750Hz allowed a sufficient suppression of relevant artifacts in the majority of scans. No strong correlation between the image quality and the distance between the cardiac silhouette and the device was detected. The presented work shows the feasibility of clinical cardiac MRI at 3.0T including GRE-CINE and wideband PSIR-LGE for patients with ICDs.

Purpose: To evaluate whether visually assessed high-quality prostate MRI, as defined by the Prostate Imaging Quality (PI-QUAL) scoring system, is sufficient to ensure consistent AI model performance, or whether hidden acquisition-related technical effects introduce variability that cannot be detected through visual inspection. Methods: This retrospective study included 568 bpMRI scans from seven institutions, classified using PI-QUAL v2 into high-quality (scores 2-3) and low-quality (score 1) groups. Quantitative image quality features were extracted using MRQy from T2w and ADC sequences. Three experiments were conducted on high-quality scans: (1) unsupervised clustering to identify hidden technical batch effects; (2) V-Net models (M) for prostate capsule segmentation on T2w using single-cluster and balanced multi-cluster training strategies; and (3) VoxelMorph-based deformable registration (M2) to align ADC maps to T2w images. Performance metrics included DSC, precision, recall, volume ratio, HD95, ASSD, and TRE. Kruskal-Wallis tests assessed cross-cluster variability, Two One-Sided Tests (TOST) evaluated performance equivalence, and bootstrap resampling quantified training efficiency. Results: Clustering identified four clusters among 277 high-quality bpMRI scans. Single-cluster models showed variable performance (DSC=0.41-0.85) with significant batch effects (p<0.001), even with 140 training samples. The balanced multi-cluster model (n=60) eliminated batch effects (p=0.075), achieved equivalent performance (p<0.001), and demonstrated 2.23-fold greater training efficiency despite using 2.3-fold fewer samples. Registration performance also varied significantly across clusters (TRE=0.86-5.47 mm, p<0.001). Conclusion: Acquisition-related batch effects significantly impact AI generalizability in prostate MRI, even among high-quality scans. Quantitative clustering enables detection of technical variations and supports diversity-aware training strategies that achieve equivalent performance with 2.3-fold fewer samples while eliminating batch effects.

Biological aging, the decline in physiological resilience that increases our vulnerability to disease and death as we grow older, can be measured from standard clinical-laboratory panels. Measurement of biological aging from laboratory data embedded in electronic health records (EHRs) could enhance precision healthcare. However, suitability of EHR-embedded laboratory records for this purpose is uncertain. We evaluated feasibility of EHR biological age measurement in a cohort of n=4,313 cardiothoracic-surgery patients at a large US Hospital and tested whether measured biological aging was associated with patient outcomes. Patients who were biologically older prior to surgery more often died or experienced adverse outcomes as compared to those measured to be biologically younger. Postoperative biological-aging trajectories were associated with length of intensive-care-unit (ICU) stays and in-hospital death. EHR laboratory data provide an opportunity for scalable implementation of biological aging measurement within healthcare systems, with potential to enhance patient monitoring and risk stratification.

Background. Real-world 12-lead electrocardiograms (ECGs) often have missing or corrupted leads. A deep-learning classifier can be prepared for this by random lead masking during training, or by injecting clinical knowledge, for example by pulling its attention towards the leads that guidelines use for each diagnosis. Whether this knowledge adds robustness beyond masking is unknown. Methods. Pre-registered comparison of three arms sharing one network (per-lead 1D-ResNet encoder and per-diagnosis attention over the 12 leads): standard training (M0); random whole-lead masking (M1); masking plus a Kullback-Leibler penalty pulling attention towards a fixed guideline-derived key-lead map (M2, guideline-based attention guidance). Six diagnoses; five seeds per arm. Development on PTB-XL (21,799 ECGs); single sealed evaluation on PTB-XL fold 10 (2,198 ECGs) and on CODE-15% (233,204 exams, external validation). Primary endpoint: AUROC difference M2 - M1, averaged over first-degree AV block, right and left bundle branch block and atrial fibrillation and over 1 to 6 lost (zeroed) leads, with a patient-level paired BCa bootstrap 95% confidence interval. Results. The guidance did not change robustness: {Delta}AUROC -0.0010 (95% CI -0.0033 to +0.0006), an interval entirely within the pre-specified +-0.02 margin; the result was unchanged with another bootstrap seed, other interval methods and leave-one-diagnosis-out analyses. Masking improved AUROC: M1 - M0 = +0.0134 (95% CI +0.0102 to +0.0172) when one key lead was lost, and the number of lost leads tolerated before AUROC fell by more than 0.02 rose from 7 (M0) to 10 (M1). M2 attention matched the map and M2 was non-inferior on intact ECGs, but its calibration slope was lower for all six diagnoses in both datasets. M2 lost slightly more when all key leads were lost together (-0.0067) and gained only when one to three leads remained (exploratory; AUROC 0.9408 vs 0.9210 with 11 leads lost). CODE-15% reproduced every pattern (primary -0.0007, 95% CI -0.0011 to -0.0003). Conclusions. Random lead masking suffices to make an ECG classifier robust to missing leads; adding guideline knowledge aligns attention with clinical criteria but adds no robustness and worsens calibration. Guideline-aligned attention should not be taken as evidence of robustness.

DNA methylation-based classification is increasingly a cornerstone of CNS tumor diagnostics. Recently,nanopore sequencing enables very rapid methylation profiling to support surgical decision making. Time-constrained sequencing yields sparse methylation profiles, necessitating specialized classifiers for accurate predictions. All publicly available sparse-profile classifiers rely on a single reference set of only 2,801 samples published in 2018 and its accompanying classification system. Here we present Sturgeon V2, trained on nearly 10,000 methylation profiles assembled from a range of public repositories and collaborations. To harmonize differences in tumor class labels, we developed an iterative semi-supervised relabeling strategy with a two-tiered concordance filter. We demonstrate that this filtering is essential for optimal performance. Compared to Sturgeon V1, we further redesigned the classifier architecture to provide more stable, and consistent confidence estimates while accumulating sequencing data and we introduce an updated class system including new classes, such as meningioma subtypes, PATZ- fusion neuroepithelial tumors and PLAG1/2-amplified embryonal tumors. In a retrospective intraoperative cohort of 94 pediatric cases, Sturgeon V2 was confidently correct in 85 cases (Sturgeon V1: 78) with one confident misclassification, and in simulated sequencing runs (n=1880) reduced the error rate among confident calls from 7.1% to 2.8%, outperforming publicly available alternatives. Sturgeon V2 is the first openly available, locally executable classifier for sparse methylation profiles trained beyond the 2018 reference era.

BACKGROUND Timely cause-of-death (COD) information guides health policy, yet many deaths in low- and middle-income countries are not medically certified. In Africa, fewer than 10% of deaths are. Verbal autopsy (VA) yields a structured symptom questionnaire and a free-text narrative. Mortality surveillance increasingly records the narrative alongside the questionnaire, but the algorithms used to produce VA-based statistics were designed for structured responses only, so narratives inform physician COD assignment but do not contribute to algorithmic COD assignment. We evaluated whether supervised models could use both modalities to improve COD assignment. METHODS We analyzed 4,999 VA records for deaths at ages 12 years or older from a national sample of registered deaths in South Africa (RSA), using physician-assigned CODs as the reference standard. We trained unimodal and multimodal pipelines on the labelled records: pretrained language models (PLMs) fine-tuned on narratives alone, machine learning (ML) models on structured responses alone, and the two combined at the data, feature, and decision levels. We also prompted a locally deployed generative large language model (LLM) on the same unimodal and multimodal inputs, without parameter updates. All were compared with InSilicoVA, the operational comparator representing current automated practice. We evaluated individual-level assignment, population-level agreement, calibration, and generalization across provinces in RSA, reporting these metrics rather than ranking pipelines on any one of them. RESULTS Every supervised pipeline improved on InSilicoVA by 8.5-15.9 percentage points in top-1 accuracy and 3.5-7.1 points in cause-specific mortality fraction (CSMF) accuracy. The narrative and structured pipelines performed similarly in aggregate but differed by cause: 63.9% of deaths received identical predictions, 11.6% were correct only from the narrative pipeline, and 13.6% only from the questionnaire pipeline, with the narrative pipeline more accurate for noncommunicable causes and the questionnaire pipeline for communicable causes. Against the unimodal approaches, the decision-fusion pipelines added 4.2-5.4 percentage points in top-1 accuracy, concentrated in the most common causes, whereas feature fusion with cross-modal attention improved the less common causes, adding 6.6 points (95% CI, 1.7-11.5) in balanced accuracy and 5.5 points (95% CI, 0.6-10.3) in macro F1. Feature fusion reached the highest CSMF accuracy (95.7%). Paired differences from the better unimodal representative ranged from -0.2 to +3.4 points, with all intervals including zero. The LLM exceeded InSilicoVA on individual-level metrics and reached the highest balanced accuracy of any pipeline (60.6%), while the supervised pipelines retained higher CSMF accuracy. Supervised pipelines had lower expected calibration error than InSilicoVA. Supervised gains held across all nine held-out provinces, including every top-1 comparison and nearly every macro F1 and CSMF comparison. CONCLUSIONS PLMs using narratives alone substantially improved on current best practice and matched ML pipelines using structured responses, while multiple multimodal strategies produced further gains in individual COD assignment. The zero-shot LLM showed particular strength on rare causes, whereas supervised pipelines better recovered the population cause distribution. These findings support incorporating narratives as a core input to automated VA systems and evaluating representative fusion strategies for the intended output. External validation across settings and age groups will support broader implementation.

Objective: To determine how cohort versus nested case-control (NCC) design affects EHR-based prediction of a rare outcome. Materials and Methods: We compared two cohort and five NCC designs for predicting early-onset gastrointestinal cancer using L1-regularized logistic regression (LASSO), XGBoost, and multilayer perceptron (MLP). Models underwent nested five-fold cross-validation with model-specific hyperparameter tuning and temporal evaluation in a held-out prospective cohort representing intended deployment. Model discrimination was assessed using area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC). Model calibration was assessed using Brier scores and calibration plots before and after recalibration with isotonic regression. Feature-ranking similarity was quantified using rank-biased overlap. Results: The cohort development datasets included 256,854-415,529 patients, with 2-yr cumulative incidence of 0.12%-0.17%, compared with approximately 9% in the 1:10 matched NCC datasets. NCC designs generally produced higher internal discrimination, but temporal performance varied by algorithm and index-date definition. Cohort designs achieved the highest temporal AUPRC for LASSO (0.068), whereas diagnosis-anchored NCC achieved the highest temporal AUPRC for XGBoost (0.086) and MLP (0.070). Temporal AUPRC generally decreased as the required lead time before diagnosis increased. Raw predicted probabilities substantially overestimated observed risks in the temporal cohort. Isotonic recalibration improved probability scaling while largely preserving discrimination, although residual miscalibration remained. Cohort-based feature rankings were more consistent across algorithms, whereas NCC-based rankings depended more strongly on algorithm and index-date anchoring. Conclusion: Study design should be selected and evaluated according to the intended clinical prediction task and deployment setting.

Objective. Temperature readings are routinely stratified by race and ethnicity, though significant group differences may reflect physiology, instrument noise, or categorization rather than biology. This study tests whether these categories partition facial infrared thermography measurements beyond individual variation and baseline device noise. Approach. A secondary analysis of thermography data (n = 951 adults) compares inner-canthus maximum (TCEmax) and full-face maximum (Tmax) temperatures across five administrative categories via one-way analysis of variance (ANOVA), mixed-effects modeling, Bland-Altman analysis against oral reference, and Monte Carlo sensitivity analysis. Main results. ANOVA returns significant group differences (p < 0.001; eta-squared = 0.04 for TCEmax, 0.03 for Tmax), but the total spread (0.2 to 0.3 degrees Celsius) is comparable to device accuracy and within-subject variability. Individual variation accounts for 95% of TCEmax variance, and adding simulated Gaussian noise to participant-level temperatures (sigma = 0.10 to 0.30 degrees Celsius) preserves the complete group-mean rank order in as few as 37.5% of iterations. Significance. Testing whether a group difference survives perturbation at the scale of instrument accuracy offers a check against treating significance alone as evidence of physiology. Here, observed group differences are small relative to measurement variability, and their physiological interpretation remains uncertain, with implications for clinical tools that treat administrative categories as biologically meaningful.

Background: The C-reactive protein-albumin-lymphocyte (CALLY) index integrates systemic inflammation, nutritional status, and immune function. We evaluated its association with survival in primary lung malignancies. Methods: Embase, PubMed, Scopus, and Web of Science were searched from inception to 18 September 2026. Risk of bias was assessed using the Quality In Prognosis Studies tool. Hazard ratios (HRs) were synthesized using random-effects models with restricted maximum likelihood estimation and Hartung-Knapp adjustment. Sensitivity analyses, subgroup analyses, and meta-regression examined robustness and heterogeneity. Results: Eleven studies contributed 13 cohorts to the overall survival (OS) analysis and 8 cohorts from seven studies to a combined recurrence/progression-free survival (RFS/PFS) analysis. Lower CALLY was associated with poorer OS (HR 1.78, 95% CI 1.49-2.12; I2=71.6%) and poorer RFS/PFS (HR 1.60, 95% CI 1.40-1.84; I2=0%). Associations persisted in leave-one-out analyses. No statistically significant subgroup differences were detected. Funnel-plot asymmetry suggested small-study effects for OS and RFS/PFS. Seven studies had moderate and four had high overall risk of bias. The certainty of evidence was rated as low for both OS and RFS/PFS, primarily due to risk of bias, heterogeneity, and endpoint indirectness. Conclusion: Lower CALLY was associated with adverse survival outcomes in the current synthesis. Methodological limitations and variation in biomarker measurement and endpoint definitions restrict interpretation. Prospective validation using standardized measurements and clinically coherent outcomes is needed to establish incremental prognostic value and clinical utility. Keywords: CALLY index; lung cancer; prognosis; overall survival; systematic review; meta-analysis

Longitudinal clinical cohorts contain heterogeneous progression patterns that are obscured by coarse diagnostic labels. Discovering these latent progression phenotypes requires models that distinguish patients not only by their outcomes, but also by how their clinical trajectories evolve. This paper proposes TRACE, an interpretable temporal phenotyping method that enables the learning of outcome-informed progression phenotypes from longitudinal trajectories. TRACE combines a trajectory- and outcome-aware contrastive objective with a probabilistic predictive phenotype layer, encouraging phenotypes to capture progression structure while remaining predictive of diagnostic state. A dual-attention mechanism further supports interpretation by identifying the visits and clinical variables most associated with each phenotype assignment. TRACE is evaluated on the Alzheimer's Disease Neuroimaging Initiative (ADNI) cohort for cognitive disease modeling, showing improved performance in both diagnostic prediction and phenotype discovery compared with other temporal phenotyping baselines. Finally, by allowing the number of phenotypes to exceed the diagnostic labels, TRACE enables discovering additional progression patterns that are not captured by labels defined a priori.

Open-loop insulin pump therapy depends on clinical expertise for dose initiation and titration. We developed a long short-term memory-based system, using 1738 treatment courses from 1641 patients with type 2 diabetes for initial prescribing and daily basal and bolus titration. Patient-grouped five-fold validation showed closer agreement with recorded prescriptions for subsequent titration than for initiation. Prospective evaluation in 97 patients from the model-directed arm of a randomized trial yielded a median derived time in range (3.9-10.0mmol/L) of 84.5%, with low hypoglycemic exposure. Favorable control extended to previously treated patients. Higher mean glucose during the first 24 hours was associated with poorer glycemic control thereafter. The association between distance from development data and glycemic outcomes was stronger when early glucose responses were included alongside baseline characteristics. These findings support model-directed pump dosing and highlight early treatment response as a clinically relevant signal for assessing applicability and identifying patients who may require additional oversight. Trial registration: ClinicalTrials.gov, NCT07066891.