Biology Methods and Protocols

Objective: To develop and validate a supervised text-embedded transformer matching model to identify fall injuries in Medicare data, and evaluate the model's performance -- alongside a validated rule-based algorithm-- against "ground truth" from an external reference standard (self-reported fall injuries leading to medical attention). Materials and Methods: Text embeddings of ICD-10-CM and CPT codes in Medicare claims/encounters from participants in the Strategies to Reduce Injuries and Develop Confidence in Elders (STRIDE) trial served as model inputs. Trained on annotated claims/encounters occurring within +/- one month of self-reported fall injuries leading to medical attention, the transformer model generated a continuous 0-1 probability that each claim/encounter was for a fall injury. The model was then applied to all claims/encounters in STRIDE and compared alongside the rule-based algorithm to the external reference standard. Results: The model achieved an area under the curve (AUC) of > 0.96 against annotated claims/encounters in 9 out of 10 holdout folds and 0.85 in the remaining fold. In the full STRIDE dataset, the model achieved a peak AUC of 0.86 (95% CI, 0.84-0.87) against the external reference standard, with results comparable to the rule-based algorithm. Discussion: Relative to rule-based approaches, which typically generate binary outcomes, the continuous event probability generated by the transformer model could support clinical endpoint adjudication, with high-probability predictions treated as events, moderate-probability predictions being adjudicated, and low-probability predictions treated as non-events. Conclusion: A text-embedded transformer model identified fall injuries with comparable accuracy to a rule-based algorithm, demonstrating "proof of concept" for use in endpoint adjudication.

Background. Comparing long-term outcomes of gastric cancer surgery between hospitals requires a common reference. Risk-adjusted benchmarks depend on the model and population used to build them and are difficult for another hospital to reproduce. TNM stage is a prognostic classification already recorded for every resected gastric cancer. We report unadjusted stage-specific overall survival from a consecutive cohort with complete follow-up as reference data and propose a minimal dataset for producing comparable figures elsewhere. Methods. The cohort comprised 690 consecutive eligible patients who underwent curative-intent D2 gastrectomy for stage I to III adenocarcinoma at Ruijin Hospital, Shanghai, in 2019. All operations were performed by eight experienced, high-volume consultant surgeons. Stage followed the AJCC 8th edition: pTNM after upfront surgery (n = 570) and ypTNM after neoadjuvant chemotherapy (n = 120). Overall survival was estimated by the Kaplan-Meier method without adjustment. Results. There were 199 deaths, 181 within five years. Every survivor was followed for at least 68.1 months. After upfront surgery, five-year overall survival was 95.0% (95% CI 91.4 to 97.1) for stage I, 83.8% (76.5 to 89.0) for stage II and 56.5% (49.2 to 63.1) for stage III, and was 95.0% in IA and 95.2% in IB, falling to 26.7% in IIIC. After neoadjuvant chemotherapy, five-year survival was 78.6%, 61.9% and 31.3% for ypStage I, II and III. Pooling pTNM and ypTNM lowered stage III five-year survival to 50.2%. The proposed minimal dataset has 16 items, 13 of them recorded during the admission for gastrectomy. Conclusions. These figures are intended as a reference for comparing hospitals rather than as a standard of care. We propose 16 items, most of them routinely recorded, from which another hospital can produce the same table and compare its results by stage.

Objective: To evaluate the case-level performance of deep-learning segmentation models for detecting extrahepatic bile duct stones on representative intraoperative cholangiography images (IOC) and to characterize the completeness of individual-stone localization. Background: Retained bile duct stones can cause biliary obstruction, cholangitis, and pancreatitis. However false-positive interpretation of filling defects may prompt additional downstream procedures. Computer vision has been applied to biliary anatomy recognition and IOC adequacy assessment, but patient-level stone detection and individual-stone localization remain insufficiently studyed. Methods: Representative IOC images were annotated for extrahepatic biliary anatomy and stones, with case-level stone status established using a composite clinical reference standard. Two deep-learning models were developed to delineate the common bile duct and common hepatic duct and to detect and localize stones. Case-level diagnostic performance was evaluated against the composite clinical reference standard, and individual-stone localization was evaluated against expert-reviewed annotations. Results: On the held-out 125 patients test set, MiT-B2-UNet identified 23 of 25 stone-positive cases and 95 of 100 stone-negative cases, corresponding to a sensitivity of 0.920, specificity of 0.950, and AUC of 0.986. nnU-Net identified 19 of 25 stone-positive cases and 98 of 100 stone-negative cases, corresponding to a sensitivity of 0.760, specificity of 0.980, and AUC of 0.959. At the individual-stone level, MiT-B2-UNet and nnU-Net localized 31 of 59 and 25 of 59 annotated stones, respectively; all annotated stones were localized in 13 of 25 and 12 of 25 stone-positive cases. Conclusions: Deep-learning models can identify stone-positive IOC cases and localize individual stones. This technology may help inte

BackgroundEvaluating fairness in artificial intelligence (AI) is essential before clinical implementation to prevent unintended harm across diverse patient populations. This study assesses performance differences of an image-based surgical site infection (SSI) detection model across age, gender, geography, race, and skin color. MethodsModel development/validation study using retrospective, multicenter data across surgical subspecialties from adults who submitted an image of their incision via the electronic health record within 30 days of their procedure between 2019-2024. Area under receiver operating curve (AUROC) was compared by age (<55 vs. [&ge;]55 years old), gender (male vs. female), geography (Rochester, Minnesota vs. Jacksonville, Florida vs. Scottsdale, Arizona vs. regional Minnesota/Wisconsin areas), race (White/Caucasian vs. non-White/Caucasian), and skin color (Fitpatrick I-III vs. IV-VI and Monk A-E vs. F-J). ResultsTraining data contained 13,702 images from 4,274 patients between 2019-2022 (50.1% [&ge;]55 years old; 63.2% female; 91.6% White/Caucasian; 58.5% Rochester, 17.3% Florida, 10.5% Arizona; 96.3% light skin tone), of which 1,348 images (9.8%) had an SSI. No differences in SSI detection existed for age, gender, race, or skin color (AUROC 0.83-0.87, p>0.09), however geographical differences were significant: lowest AUROC 0.83 (95% CI 0.77-0.88) vs. highest AUROC 0.87 (95% CI 0.84-0.89, p=0.01). Temporal validation using 4,242 images from 1,353 patients between 2022-2023 showed intact fairness across subgroups ({Delta}AUROC [&le;]0.06) except geography ({Delta}AUROC [&le;]0.16). ConclusionsThe model behaved fairly across age, gender, race, and skin color, however geographical differences persisted. These findings have practical implications underscoring the importance of evaluating bias in clinical AI before deployment.

Accurate classification of lung adenocarcinoma growth patterns from hematoxylin and eosin (H\&E)-stained whole-slide images is essential for treatment planning and prognosis. We develop and evaluate an exponentially weighted ensemble deep learning approach that combines five architectures (EfficientNet-B3, DeiT3, Swin Transformer, Vision Transformer, ConvNeXt) with weights determined by a softmax transformation of per-model Quadratic Weighted Kappa (QWK) scores. Using 25,545 image patches at 40x magnification from the Dartmouth-Hitchcock dataset, we evaluated performance with stratified 5-fold cross-validation. Individual model performance ranged from Cohen's kappa kappa = 0.5563 (ConvNeXt) to kappa = 0.9561 (EfficientNet-B3). The ensemble achieved 96.52% accuracy and kappa = 0.9648, exceeding the best individual model by delta kappa = +0.0087 with statistical significance (McNemar's test, chi^2 = 50.84, p < 10^-12; bootstrap 95% CI on delta kappa [+0.0056, +0.0117]. Pairwise disagreement rates of 16.6--43.2% between architectures indicate that the models capture different aspects of tissue morphology, and the framework automatically down-weights unreliable components, as illustrated by ConvNeXt receiving only 5.7% ensemble weight despite a single-fold training failure. The approach offers a clinically grounded method for ensemble aggregation in digital pathology.

Background Existing insulin resistance (IR) indices are predominantly developed in diabetic cohorts, limiting their generalizability. We developed a novel deep neural network-derived IR index (DNN-IR) using a Mixture-of-Experts (MoE) framework and evaluated its predictive performance for incident cardiovascular disease (CVD) and mortality in general populations. Methods We utilized data from three cohorts: the cross-sectional REACTION study (Fujian subcohort, 2011-2012) for DNN-IR derivation and internal validation; and two prospective cohorts, NHANES (1999-2018, linked to the National Death Index) and CHARLS (2011-2018), for external validation. The DNN-IR was developed using a deep learning model based on a Mixture-of-Experts (MoE) architecture, trained on the REACTION dataset. We evaluated the DNN-IR's utility in predicting incident CVD, cardiovascular mortality, and non-cardiovascular mortality among 13,889 NHANES and 7,047 CHARLS participants. Predictive performance was assessed via the area under the receiver operating characteristic curve (AUC). Multivariable logistic regression, restricted cubic splines, and Kaplan-Meier analyses characterized the associations between DNN-IR and clinical outcomes. Results In the REACTION cohort, DNN-IR demonstrated superior predictive performance for atherosclerotic outcomes, achieving AUROCs of 0.89 (training) and 0.84 (internal validation). In the external CHARLS cohort (median follow-up: 7 years; 1,135 incident CVD cases [16.1%]), DNN-IR yielded AUROCs of 0.72 for incident CVD and 0.77 for all-cause mortality. Fully adjusted models showed that each 1-SD increment in DNN-IR was associated with a 23% higher CVD risk (OR=1.23, 95% CI: 1.14-1.32), exhibiting a predominantly linear dose-response relationship (P-nonlinearity=0.453). In NHANES, DNN-IR robustly predicted cardiovascular (AUROC=0.77) and all-cause mortality (AUROC=0.72), alongside specific mortalities like diabetes (0.91), Alzheimer's disease (0.88), and kidney disease (0.96). Higher DNN-IR levels correlated with stepwise increases in cumulative mortality (log-rank P<0.001). Conclusions The MoE-derived DNN-IR index demonstrated robust and stable performance in predicting atherosclerosis, incident CVD, cardiovascular mortality, and all-cause mortality in the general population. Further validation in larger, more diverse cohorts is warranted to support its broad clinical applicability.

To explore adapting qPCR systems for end-point nucleic acid quantification using dyes such as SYTO-9, we quantified serial dilutions of DNA and RNA standards in the range of 0.75 - 200 ng/{micro}l on 384-well qPCR devices. SYTO-9 fluorescence was successfully measured using standard SYBR Green settings. Blank-subtracted relative SYTO-9 signal showed a logarithmic dependence on DNA/RNA concentration (R2 > 0.95). Measurements were highly stable with different incubation times, temperatures of up to 95{degrees}C, and photobleaching. The described approach is a valuable QC option for high-throughput DNA/RNA isolations and could be adapted to additional fluorometric assays beyond nucleic acids.

Purpose: To evaluate whether fluorescence lifetime imaging ophthalmoscopy (FLIO) combined with deep learning can detect metabolic signatures for classification of type 2 diabetes mellitus (T2DM). Design: Cross-sectional analysis of participants included AI-READI dataset (version 3) with FLIO imaging and and hemoglobin A1c (HbA1c) measurement. Subjects: 1,783 participants from the AI-READI dataset (version 3) with HbA1c measurements and FLIO imaging scans (6,912 total): 671 normoglycemic, 726 prediabetic, and 386 diabetic. Methods: Mean fluorescence lifetime maps were generated using a center-of-mass approach and used as inputs to AI models. We trained convolutional neural networks (CNNs), ResNet-18, and XGBoost under three-class (normal, prediabetic, diabetic) and two binary (normal vs. impaired; normal vs. diabetic) classification schemes, using nested 5-fold cross-validation with participant-level grouping. Main Outcome Measures: Macro-averaged accuracy, F1 score, area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, and positive predictive value (PPV). Results: Group-averaged lifetime maps demonstrated consistent spatial differences across glycemic groups, with progressively longer lifetimes from normal to diabetic participants. The CNN achieved the best overall performance in the 3-class classification (accuracy 0.41 +/- 0.03, F1 score 0.39 +/- 0.02, AUROC 0.58 +/- 0.02), compared to the random classifier for 3-class classification (AUROC = 0.50; accuracy = F1 = 0.33). ResNet-18 and XGBoost showed similar performance (AUROC 0.53-0.58). Confusion matrices revealed substantial overlap between classes, with frequent misclassification toward the prediabetes group. Binary reformulation (normal vs. diabetic) improved performance substantially, with the CNN resulting in AUROC 0.63 +/- 0.02 and XGBoost 0.67 +/- 0.07. Conclusions: FLIO-derived lifetime maps capture metabolic signals associated with glycemic status but yield modest classification performance with current AI models. These findings highlight both the potential and the challenges of using FLIO for early metabolic screening and monitoring, informing future development of clinically applicable imaging biomarkers.

Machine learning workflows frequently incorporate data preprocessing to enhance predictive performance. However, the need for Super Learner ensembles made up only of preprocessing-invariant tree-based algorithms remains unexplored. Using three benchmark clinical classification datasets, this study examined how preprocessing affected the Super Learner's prediction performance, learner weight distribution, and oracle behavior. The Heart Disease (207 observations), Indian Liver Patient Dataset (583 observations), and Pima Indians Diabetes (768 observations) datasets were used to create a Super Learner ensemble model that included Classification and Regression Trees (CART), Random Forest, Ranger, and Extreme Gradient Boosting (XGBoost). Models were evaluated under raw and preprocessed data conditions using repeated cross-validation. Predictive performance was assessed using the area under the receiver operating characteristic curve (AUC), Matthews correlation coefficient (MCC), and Brier score. Learner weight allocation and Oracle Gap were compared using paired Wilcoxon signed-rank tests with Benjamini-Hochberg adjustment. Preprocessing produced negligible changes in predictive performance for the Heart Disease and Pima datasets. For the ILPD dataset, preprocessing significantly improved AUC (0.746 to 0.752; adjusted p = 0.0017) and reduced the Brier score (0.177 to 0.175; adjusted p < 0.001). Learner weights remained largely stable, although Random Forest replaced Ranger as the dominant learner for the Heart Disease dataset. Oracle Gaps remained extremely small (<0.002) across all datasets and did not differ significantly between preprocessing conditions. Preprocessing provides limited benefit for Super Learner ensembles composed of preprocessing-invariant learners and does not materially alter their oracle behavior. Preprocessing decisions should therefore be guided by dataset characteristics rather than adopted as a universal modelling practice.

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 [-&gt;] LLM [-&gt;] 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.

Context Current guidelines for primary aldosteronism (PA) recommend screening all patients with hypertension. Given low screening rates, an automated artificial intelligence (AI)-based risk-stratified approach may facilitate guideline implementation. Objective To develop an AI-based prescreening tool using electronic health record (EHR) data to identify at-risk patients. Design Model development retrospective cohort study. Ensemble architecture using eXtreme Gradient Boosting, random forest, and extremely randomized trees was trained on EHR data (1986-2025) and evaluated on a test set of patients with hypertension. Setting Single institution system spanning community clinics and tertiary/quaternary centers reported via Mayo Clinic Platform. Patients Adults with hypertension, International Classification of Disease (ICD) diagnosis of PA, or had plasma renin/aldosterone testing. Main Outcome Measures Predicting PA diagnosis (disease) or negative workup (control) using age, sex, ICD diagnoses, vitals, labs, and medications by area under receiver operating characteristic (AUROC), estimated calibration error (ECE), and specificity. Results Of 22,264 patients (1,833 disease and 20,431 controls), median time from initial encounter to screening was 5.0 years (IQR 0.5-12.6). The model demonstrated AUROC 0.709 (95% CI 0.688-0.730) with ECE 0.037. At a 0.5 cut-off threshold, the model had high specificity (71.6%) and low false positivity (28.4%). In a test cohort of 1,279,455 adults with hypertension, the model identified 5,295 (0.4%) for first-pass screening. Conclusions A high-specificity model can provide a pre-screening PA pathway using routine EHR data to identify and prioritize patients for PA screening. This AI-based approach can risk-stratify patients, guide sequencing of testing/referrals, and pragmatically facilitate adoption of current guidelines into practice.

Background Hepatocellular carcinoma (HCC) treatment selection demands nuanced integration of heterogeneous patient data, yet prevailing predictive models rely on restricted data modalities and oversimplified therapeutic frameworks, compromising clinical translation. Objective We developed and validated a multimodal artificial intelligence framework to guide optimal treatment strategy selection across the full spectrum of HCC interventions. Methods This retrospective study comprised 1,043 HCC patients (development cohort, January 2017-December 2023) and 55 external validation patients (2023) from Wuxi Peoples Hospital. We engineered Embedding-Augmented Extra Trees (ET-Emb), a novel model fusing structured clinical variables with contextual text embeddings derived from medical histories and radiology reports. ET-Emb quantifies probabilities for five primary treatments: open/laparoscopic resection, transarterial chemoembolization, radiofrequency ablation (RFA), and chemotherapy. Model performance was rigorously assessed via 10-fold cross-validation and external validation using ROC-AUC and PR-AUC metrics. Results ET-Emb demonstrated robust performance in the development cohort (ROC-AUC: 0.84 {+/-} 0.04; PR-AUC: 0.55 {+/-} 0.06), significantly outperforming established benchmarks. This generalizability was preserved in external validation (ROC-AUC: 0.77 {+/-} 0.02; PR-AUC: 0.47 {+/-} 0.03). SHAP analysis identified textual clinical narratives and socioeconomic determinants as critical predictive drivers. Conclusions By unifying structured and unstructured data modalities, ET-Emb delivers accurate, multi-treatment strategy prediction for HCC. Its clinical validity and the demonstrated significance of textual features establish multimodal AI as an essential paradigm for simulating complex oncological decision-making, positioning ET-Emb as a transformative tool for precision HCC management.

Developments in Nanopore sequencing have enabled telomere to telomere genomic assembly as a routine technique in genomic research. Nanopore DNA sequencing for genomic assembly is typically performed on native DNA molecules, making it particularly sensitive to the quality of input DNA, with contaminating molecules limiting data yields and reducing read quality. As pangenome analysis gains interest, particularly in non-model plant species which are often rich in inhibitory secondary metabolites, the development of methods which can improve the quality and throughput of nanopore sequencing is essential. Here we describe a method for isolation of total DNA from the leaf tissues of diverse Viridiplantae species. The initial lysis buffer consists of a modified CTAB buffer, incorporating dimethyl sulfoxide for the reduction of viscosity, which can be problematic in many plant DNA preparations. An organic extraction with 2-butoxyethanol is utilized to further extract phenolic compounds which may be sufficiently hydrophilic to evade chloroform extraction, while reducing aqueous phase volume. Further cleanup via cesium chloride (CsCl) ultracentrifugation is performed to minimize the carryover of residual contaminating macromolecules. Samples prepared using this method are of consistent high quality, even when extracted from challenging late season leaf tissue or secondary metabolite rich species. Sequencing results from samples prepared by this method outperform those obtained from typical modified CTAB DNA isolation techniques in both quantity and quality. We tested sequencing performance from Vitis DNA isolated using a modified CTAB method and Vitis DNA isolated using the CsCl ultracentrifugation-based method described here. DNA isolated via the method described here produced 83% more >Q10 sequence data (52.61 Gb vs. 28.8 Gb), resulted in a 60% greater read N50 despite more handling steps (32.78kb vs. 20.45kb), and resulted in a higher modal read quality (Q27 vs. Q24). The consistency of this method across diverse plant taxa suggests its use as a general method for DNA isolation prior to Nanopore sequencing and genomic assembly for diverse plant taxa.

Background: Robust evidence linking individual surgeon technical performance to patient outcomes in complex minimally invasive surgery remains scarce. Crowd-sourced video assessment offers a scalable alternative to expert review, but its clinical relevance in complex oncologic procedures such as robot-assisted minimally invasive esophagectomy (RAMIE) is unestablished. Methods: In this exploratory study, video recordings of 60 consecutive RAMIE procedures performed by 7 surgeons (2018-2023) were segmented into five representative surgical steps and rated by 93 blinded crowd-raters per segment using a modified GEARS instrument. Tertiles were used to compare top-performing operations and steps against the remaining two terciles for overall and step-specific performance, and associations with perioperative outcomes were analyzed. Results: Top-tercile overall performance was associated with significantly shorter operative time (abdominal, thoracic, and total) and less intraoperative blood loss (all p<0.05), without differences in postoperative outcomes. In exploratory step-specific analyses, top-tertile performance during infracarinal lymphadenectomy was associated with significantly fewer surgical complications (30.0% vs 60.0%; p=0.028), fewer severe surgical complications (25.0% vs 52.5%; p=0.043), and higher lymph node yield (22.8 vs 18.9; p=0.037), while the purse-string suture step showed the broadest operative-time benefit. These step-specific findings are hypothesis-generating and were not adjusted for multiple testing. Conclusion: Crowd-sourced video assessment enables scalable, step-specific evaluation of surgical performance in RAMIE associated with intraoperative blood loss and operative time. Exploratory analyses suggest that the relationship between technical performance and clinical outcomes might be step-specific, with infracarinal lymphadenectomy emerging as risky step linked to postoperative morbidity. These findings support further investigation of individualized performance assessment for quality improvement and risk stratification.

Background: Machine Learning (ML) has shown promise in histopathological diagnosis, but rare tumours, like duodenal epithelial neoplasms, are challenging because large, labelled datasets are difficult to assemble. We propose a data-efficient ML pipeline for detecting duodenal epithelial neoplasia in whole-slide images. Methods: We assembled a five-hospital dataset of haematoxylin and eosin-stained duodenal biopsy whole slide images (WSIs) at 10x magnification, comprising 8,027 normal and 401 neoplastic biopsies, including adenomas, carcinomas and neuroendocrine tumours. WSIs were divided into patches and encoded using the Hibou-B foundation model. We compared global standardisation with unsupervised site-specific standardisation of patch embeddings. Our pipeline combined principal component analysis, a Gaussian mixture model and a Random Forest classifier. We assessed generalisability using leave-one-hospital-out cross-validation across four hospitals and on an independent test set from a fifth hospital. Results: In leave-one-hospital-out cross-validation, the pipeline achieved a mean accuracy of 90%, sensitivity of 96%, specificity of 83% and an AUC of 0.96. On the external test hospital, it achieved 94% accuracy, 93% sensitivity, 95% specificity, and AUC of 0.97. Site-specific standardisation improved mean cross-validation accuracy by 20 percentage points compared with global standardisation, while the Gaussian mixture pipeline improved accuracy by 12 percentage points compared with mean pooling of patch embeddings. Conclusions: We present a data-efficient ML approach for detecting duodenal epithelial neoplasia from WSIs. Strong performance was maintained across hospitals and on an independent external test cohort despite the limited number of neoplastic training cases, supporting the feasibility of foundation-model-based approaches for rare histopathological diagnoses.

Background: Biomedical studies increasingly collect repeated measurements, such as clinical ratings, speech measures, gait summaries, and wearable-sensor features, at irregular subject-specific follow-up times. These data are often sparse and unaligned, making them difficult to use with standard classifiers that require fixed-length input vectors. Parkinson's disease provides a motivating example because progression is longitudinal and heterogeneous, but clinical follow-up is rarely observed on a common schedule. Objective: This study develops and evaluates a multiclass likelihood contrast classifier for sparse longitudinal trajectory recovery. The goal is to classify subjects using their observed repeated measurements while avoiding aggressive reduction of trajectories to averages or simple slopes. Methods: The proposed method fits one class-specific mixed-effects model per outcome class and assigns each held-out subject to the class with the largest marginal likelihood score. The main specification uses a quadratic mean trajectory with subject-specific random intercept and random slope terms. Performance is evaluated in a three-class simulation with imbalanced class sizes and in a Parkinson's motor-UPDRS sparse trajectory recovery experiment. In the Parkinson's experiment, dense histories are first used to define mathematical trajectory phenotypes; most observations are then hidden, and models are asked to recover the phenotype from the sparse record. Baselines include a linear mixed-likelihood classifier, functional k-nearest neighbors, random forest, and support vector machine. Results: In the Parkinson's sparse recovery experiment, after approximately 90% of dense observations were hidden, the proposed model achieved accuracy 0.8095, macro-F1 0.8160, MCC 0.7107, and macro-AUC 0.9268. A train-validation-test diagnostic gave similar held-out behavior, with validation macro-F1 0.7980 and test macro-F1 0.8160 for the proposed model. Conclusion: Multiclass likelihood contrasts offer an interpretable and competitive approach for sparse, unaligned longitudinal classification. The Parkinson's analysis should be read as an algorithmic sparse trajectory recovery study rather than clinical diagnostic validation. Future work should evaluate the method with externally assigned clinical outcomes, multivariate longitudinal biomarkers, prospective cohorts, and ordinal generalized mixed-model likelihoods that better match bounded clinical rating scales.

Background and Objectives: Robotic-assisted natural orifice specimen extraction surgery (NOSES) is a minimally invasive approach for mid-rectal cancer, but it is technically complex and lacks a standardized operational framework. This study aimed to propose and preliminarily validate a structured modular robotic NOSES II surgical procedure for mid-rectal cancer. Methods: This was a retrospective observational study that consecutively enrolled 11 patients with mid-rectal cancer who underwent modular robotic NOSES-II surgery at our center between December 2024 and August 2025. All procedures were performed by the same surgical team strictly following the predefined six-module structured surgical protocol. Perioperative indicators, pathological outcomes, and patient-reported outcomes (PROs) at 4 to 8 months postoperatively were collected. Key techniques were analyzed via high-definition surgical videos. Results: All 11 procedures were completed successfully without conversion to open surgery. The mean operative time was 299.6 plus or minus 54.2 minutes, and the mean intraoperative blood loss was 94.1 plus or minus50.6 mL. The mean number of harvested lymph nodes was 17.4 plus or minus 8.4, with a 100% R0 resection rate. No severe complications (Clavien-Dindo grade [&ge;] III) occurred. The mean postoperative hospital stay was 8.2 plus or minus 1.5 days. Postoperative PROs indicated good defecatory function, urinary function, and overall quality of life. Conclusions: Preliminary findings demonstrate that the structured modular robotic NOSES-II procedure is safe and feasible for the treatment of mid-rectal cancer. This modular protocol provides a clear and reproducible technical framework for this complex procedure, and it is expected to achieve favorable functional preservation while ensuring oncological radicality.

Motivation: Graph Neural Networks (GNNs) have gained increasing interest in the biomedical domain, as the integration of prior knowledge and deep neural networks has the potential to enhance insights into molecular processes and disease mechanisms. However, a comprehensive and systematic assessment of model architectures, data modalities, graph structures, and their performance for graph signal classification in the biomedical domain is yet to be performed. In order to close this gap, we conducted a benchmarking study on multiple GNNs on a Protein-Protein Interaction (PPI) network for Kidney Renal Clear Cell Carcinoma and Breast cancer subtype prediction, performing an in-depth investigation of architectures, incorporating skip connections and various data modalities. Results: While none of the GNNs outperforms the structure-agnostic Multi-Layer Perceptron baseline, all of them can handle bimodal data (gene methylation and expression) and offer the ability to gain explainability based on PPIs. We offer practical guidelines for applying GNNs to graph signal processing tasks specifically for cancer classification. Depending on the underlying dataset and PPI structure employed, models on different data modalities outperform others. Overall, we suggest using ChebNet, which tends to outperform the Graph Convolutional Network and the Graph Attention Network in cancer subtype prediction. We recommend using GNN architectures that employ a simple flattening readout layer, as they provide better classification performance and faster training time than those with global average pooling. Additionally, we tested residual connections, but they had only an insignificant impact on classification performance.

Purpose The Karolinska Liver Metastases (KaroLiver) cohort was established to investigate associations between clinical characteristics and histopathological features in patients treated with curative intent for colorectal cancer liver metastases (CRLM). The cohort combines whole-slide digital histopathology images with detailed oncological, surgical, radiological and survival data, enabling comprehensive analyses of treatment trajectories, clinical outcomes and metastatic tumour biology. Participants KaroLiver is a retrospective observational cohort comprising all consecutive patients who received curative-intent, liver-directed treatment for CRLM at Karolinska University Hospital in Stockholm, Sweden. The hospital is the primary regional referral centre for HPB surgery, serving the population of approximately 2.5 million people in the Stockholm-Gotland healthcare region. Patient enrolment is continuously updated in accordance with amended ethical approvals and evolving scientific questions. The cohort currently comprises 811 patients who underwent 1204 liver interventions between February 2012 and January 2022. Detailed clinical, oncological, surgical, pathological, molecular, recurrence and survival data are collected. Findings to date Median overall survival (OS) in the current cohort is 51.0 months (95% CI 46.2-57.1 months), and median recurrence-free survival (RFS) is 10.4 months (95% CI 9.2-11.9 months). The five-year OS rate is 44.9% (95% CI 41.2-48.8%). Studies using the cohort have so far identified a liver injury-derived stromal capsule in a subset of metastases, associated with improved survival. The cohort has also enabled the identification of histopathological markers of tumour biology, sex-based differences in treatment and survival, and associations between post-hepatectomy liver failure and oncological outcomes. Future plans Current research priorities include advanced histology-based prognostic scoring, sex differences in recurrence and retreatment, tumour biology and outcomes in early-onset versus average-onset CRLM, as well as CT- and MRI-based radiomics, all integrated within KaroLiver's histopathological framework. Data sharing is supported, given that regulatory requirements are met. Retrospective accrual and outcome updates will continue for current and future studies, subject to the required approvals.

Background: Clinicopathologic (SARCULATOR) and transcriptomic (CINSARC) models are established prognostication tools in soft tissue sarcoma, but neither directly represent the sequence context of tumor genomic alterations. Evo2 is a genomic foundation model trained to learn the grammar of DNA/RNA sequences. We tested whether Evo2-derived features from tumor short variants, fusions, and rearrangements improve survival prognostication beyond clinicopathologic and transcriptomic risk models. Methods: We analysed formalin-fixed, paraffin-embedded tumor material and clinical data from 102 patients with dedifferentiated liposarcoma (n = 25), high-grade leiomyosarcoma (n = 25), and undifferentiated pleomorphic sarcoma (n = 52). A separate TCGA-SARC cohort of comparable histologies (n = 129) served as external validation. Reconstructable alterations (short variants, fusions and rearrangements) were converted into sequence inputs, embedded with Evo2 and aggregated into patient-level genomic representations. Cox proportional hazards models incorporated genomic features alone (Evo2) or together with clinicopathologic (SARCULATOR) or transcriptomic (CINSARC) predictors across overall survival (OS), progression-free survival (PFS), disease-free survival (DFS), relapse-free survival (RFS), and metastasis-free survival (MFS). Discrimination was assessed using Harrell's concordance index (C-index). Results: Evo2 features were generated from 174 study-cohort sequences and 7298 TCGA-SARC sequences. In the study cohort, adding Evo2 to SARCULATOR increased the reported C-index for OS from 0.620 to 0.770, PFS from 0.625 to 0.755, DFS from 0.612 to 0.714, and RFS from 0.633 to 0.771. In TCGA-SARC, SARCULATOR plus Evo2 increased the OS C-index from 0.595 to 0.716 and the MFS C-index from 0.548 to 0.667. Also in TCGA-SARC, adding Evo2 to CINSARC increased the OS C-index from 0.476 to 0.653 and the MFS C-index from 0.553 to 0.650. These are descriptive point-estimate comparisons. Moreover, in TCGA-SARC Evo2 alone outperformed SARCULATOR alone and CINSARC alone, supporting independent prognostic information within altered sequence-context representations. Conclusions: Evo2-derived genomic representations improved clinicopathologic and transcriptomic prognostication across sarcoma cohorts and survival endpoints. These findings support genomic foundation models as an independent and complementary layer for sarcoma risk stratification through direct modeling of altered tumor sequence biology. Confirmation of model transportability, calibration and clinical utility is required before clinical application.