Computational and Structural Biotechnology Journal

ObjectivePreterm birth (PTB) is a leading cause of neonatal morbidity and mortality worldwide, with India alone contributing nearly 27% of the global PTB burden. Although alterations in the vaginal microbiome have been implicated in PTB, its association in the Indian context is underexplored. This study aimed to investigate the association of vaginal microbiome and PTB in Indian women at the time of delivery. Study designThe vaginal swabs were collected at the time of delivery from 72 women (31 term, 41 preterm) admitted to a tertiary care hospital in Western India. Microbial DNA was extracted, and the V3-V4 region of the 16S rRNA gene was sequenced. Community composition, alpha and beta diversity, and differential taxonomic abundance were assessed using bioinformatics pipelines. ResultsAt the time of delivery, there were no significant differences in alpha or beta diversity between term and preterm groups. Principal coordinate and unsupervised clustering analyses showed no group-wise segregation. The relative abundance of individual Lactobacillus species, including L. iners and L. helveticus, did not differ significantly between the two groups. However, a modest difference in the relative abundance of Streptococcus was observed between the two groups after adjustment. ConclusionThis study found no major microbial shifts in the vaginal microbiome associated with preterm birth in this cross sectional cohort of Indian women, suggesting that vaginal dysbiosis at the time of delivery may not be a principal driver of PTB in this population. These findings underscore the need for larger, longitudinal, and ethnically diverse studies using standardized methodologies better to understand the microbiomes role in PTB risk.

Osteoporosis is a silent yet debilitating disease that often remains undetected until fractures occur. While early prediction is crucial, most studies combine male and female datasets to train a single model, introducing bias since osteoporosis risk and progression differ by gender. This study aims to develop gender-specific machine learning models that leverage longitudinal data to predict osteoporosis risk, providing tailored insights for men and women. Data were obtained from two large longitudinal cohorts: the Study of Osteoporotic Fractures (SOF) for women and the Osteoporotic Fractures in Men Study (MrOS) for men. Multiple ML algorithms were trained and evaluated for each sex, with model performance assessed using the area under the receiver operating characteristic curve (AUC-ROC). Among the tested models, the XGBoost model demonstrated the best performance for women, achieving an AUC-ROC of 0.93 using SOF data. For men, the Random Forest model achieved an AUC-ROC of 0.89 using MrOS data. Feature importance analysis identified sex-specific osteoporosis risk factors, underscoring the need for tailored prediction and management. By revealing male and female risk factors and reducing bias from combined datasets, the work advances personalized care and supports earlier, effective clinical intervention to prevent fractures and improve health outcomes.

Background & AimsThe increasing global prevalence of metabolic dysfunction-associated steatotic liver disease (MASLD) including metabolic dysfunction-associated steatohepatitis (MASH) creates an urgent need for objective methods of histopathological assessment. Conventional histological approaches are time-consuming and rely on interpreters experience. Therefore, the results obtained may suffer from high variability and only offer coarse categorisation. In this study, we propose a fully automated, deep-learning-based pipeline for the segmentation and characterisation of histological liver features for MASH/MASLD assessment. MethodsSegmentation was applied to H&E sections from 45 mice and 44 humans with MASH/MASLD. The method, which we named qHisto (quantitative histology), utilises the nnU-Net framework and quantifies key histological components of the MASH score, including macro- and microvesicular steatosis, fibrosis, inflammation, hepatocellular ballooning and glycogenated nuclei. Additionally, we characterized the tissue using novel features that are inaccessible through manual histology, such as the distribution of fat droplet sizes, aspect ratio of nuclei and heatmaps. ResultsqHisto parameters showed strong positive correlations with conventional histology scores (fat area R=0.91, inflammation density R=0.7, ballooning density R=0.49) and also with quantitative magnetic resonance imaging (fat area vs. hepatic fat fraction R=0.87). Our novel scores showed that deformation of nuclei is driven by large fat droplets rather than the overall amount of fat. ConclusionsA key advantage of our method is spatially resolved, precise histological quantification. These features provide a finely resolved assessment of disease severity than conventional categorical scoring. By automating time-consuming and repetitive readouts, qHisto improves standardisation and reproducibility of MASH/MASLD feature quantification and provides scalable, slide-wide readouts that can support histopathologists and enhance clinical assessment and therapeutic development. Impact and ImplicationsThe proposed method provides an objective, automatic tool for comprehensive, histological liver analysis of MASH/MASLD, which can be extended to other diseases and organs. By offering classic and novel quantitative parameters and scores, our method could support histologists in their daily routines and provide researchers with further insight into steatotic liver diseases.

Drug-induced liver injury (DILI) is an acute inflammatory liver disease caused not only by prescription and over-the-counter medications but also by health foods and dietary supplements. Typically, DILI patients recover once the causative substance is identified and discontinued. In contrast, autoimmune hepatitis (AIH) results from the immune-mediated destruction of hepatocytes due to a breakdown of self-tolerance mechanisms. Patients presenting with acute-onset AIH often lack characteristic clinical features, such as autoantibodies, and require prompt steroid treatment to prevent progression to liver failure. Liver biopsy currently remains the gold standard to differentiate acute DILI from AIH; however, general pathologists face significant diagnostic challenges due to overlapping histopathological features. This study integrates pathology expertise with deep learning-based artificial intelligence (AI) to differentiate DILI from AIH using histopathological images. Our AI model demonstrates promising classification accuracy (Accuracy 74%, AUC 0.81). This paper presents a detailed pathological analysis alongside AI methods, discusses the current model performance and limitations, and proposes directions for future improvements.

ObjectivesGiven the widespread use of period-tracking applications and evidence that some users rely on fertile-window predictions for pregnancy prevention, we aimed to quantify pregnancy risk arising from misclassification of biologically fertile days by period-tracking applications, and to compare this risk across calendar-based and basal body temperature (BBT)-supported period tracking and a digital contraceptive regulated as a medical device. MethodsWe conducted an observational analysis of cycles of mobile fertility application users who logged urinary luteinizing hormone (LH) tests. Biologically fertile days were defined using an LH-based reference fertile window (days -5 to 0 relative to ovulation). Three approaches were evaluated: a calendar-based period tracking application, a BBT-supported period tracking application, and a FDA-cleared digital contraceptive. Outcomes included day-specific frequency of fertile days misclassified as safe, cycle-level misclassification, and predicted pregnancy risk per cycle. Analyses were repeated in a subgroup of irregular cycles. Results543,167 menstrual cycles with a clear LH surge signature were included in the analysis. Calendar-based period tracking frequently misclassified fertile days as safe, with 67% of cycles containing at least one at-risk day and 25% containing at least one high-risk day. The mean predicted pregnancy risk per cycle was 22%, increasing to 65% in irregular cycles. BBT-supported period tracking reduced misclassification but remained associated with substantial risk (41% of cycles with at least one at-risk day; mean predicted pregnancy risk 9%). In contrast, the digital contraceptive showed consistently low misclassification (3% of cycles with any at-risk day and a mean predicted pregnancy risk of 0.5%). ConclusionsBoth calendar-based and BBT-supported period-tracking applications not intended for contraception frequently misclassify biologically fertile days and should not be considered reliable tools for pregnancy prevention. Regulated digital contraceptives demonstrate substantially lower pregnancy risk. Short condensationPeriod-tracking apps frequently misclassify fertile days as safe, including days with high pregnancy risk. In a large real-world analysis, both calendar- and BBT-supported trackers showed substantial risk, unlike digital contraception methods regulated as a medical device.

BackgroundReverse total shoulder arthroplasty (rTSA) is an increasingly common surgical procedure often performed to treat pain related to glenohumeral osteoarthritis or to rotator cuff arthropathy. Although surgical outcomes are generally excellent, recent evidence has found that postoperative pain ([≥] 3/10) two years following surgery is reported by an estimated 18% of patients. Recently, the NIH Acute-to-Chronic Pain Signatures program recommended longitudinal studies using select biomarkers to describe and predict individual patient responses to surgery. These data are not yet available for rTSA procedures. MethodsThis was a longitudinal cohort study performed at a single academic medical center. Twenty participants undergoing rTSA surgery were included, recruited from a tertiary hospital system in the southern United States. The first objective of this study was to describe changes in general pain intensity (Numerical Pain Rating Scale), widespread body pain, anxiety (General Anxiety Disorder-7), depression (Patient Health Questionnaire-9), neuropathic pain symptoms (painDETECT), and quantitative sensory testing from baseline to 6 weeks following rTSA. The second objective was to identify the baseline demographic and pain-related factors associated with 6-week postsurgical improvements in pain intensity. ResultsFrom before to after surgery, our cohort demonstrated significant improvement in shoulder pain intensity, widespread body pain, PainDETECT score, and temporal summation magnitude measured at the surgical deltoid. Degree of 6-week pain intensity improvement was associated with baseline pain intensity (F=18.79, p=0.0004) and temporal summation magnitude of the tibialis anterior (F=5.06, p=0.0380). ConclusionsPain intensity, location, nature, and mechanism can serve as biomarkers of the short-term postsurgical changes that can be expected following rTSA. Baseline pain intensity and temporal summation magnitude of the tibialis anterior were associated with the degree of pain improvement, suggesting their use for preoperative risk assessment. Future research should evaluate whether these 6-week biomarker changes are associated with the development of chronic postoperative pain at longer durations after surgery. Level of EvidenceLevel I, Prognostic Study

ObjectivesTo assess if maternal stress is higher in pregnancies with congenital heart disease (CHD) compared to low-risk pregnancies and if maternal stress is associated with placental microstructure and function. To explore if CHD alters the relationship between maternal stress and placental measures. MethodsIn this prospective observational study, 27 participants carrying a fetus with CHD and 42 participants with typical low-risk pregnancies underwent 1-2 combined diffusion{square}T2* relaxation placental MRIs from 20 weeks gestation (GA) and completed the Edinburgh Postnatal Depression Scale and State Trait Anxiety Inventory [43 male fetuses, median (IQR) GA at assessment 30.86 weeks (27.43-34.00), interval between assessments 6.00 weeks (4.86-7.14)]. 98 complete placental MRI and maternal stress datasets were available. Generalized Estimating Equations were used for analyses. ResultsHigher trait anxiety was associated with higher placental apparent diffusion coefficient (p=0.023) adjusting for CHD, sex, GA at assessment, GA at assessment2, state anxiety, depressive symptoms and previous mental health treatment. Maternal state anxiety (p=0.005) and depressive symptoms (p=0.046) were higher in pregnancies with CHD adjusting for GA at assessment and previous mental health treatment. CHD did not alter these relationships (p>0.119). ConclusionsMaternal proneness to anxiety, measured with the trait anxiety inventory, is associated with increased diffusivity in the placenta, which may reflect altered microstructural maturation. Mothers with fetal CHD show more depressive symptoms and feelings of anxiety and may benefit from screening for elevated maternal stress. The findings contribute to a growing body of research regarding the influence of prenatal stress on placental development. HighlightsO_LIMaternal stress and placental MRI data acquired in pregnancies with and without CHD C_LIO_LIMaternal trait anxiety is associated with increased placental diffusivity C_LIO_LIMaternal state anxiety and depressive symptoms are higher in fetal CHD C_LIO_LIState anxiety and depressive symptoms not associated with placental MRI measures C_LIO_LICHD did not moderate relationships between placental MRI measures and stress C_LI

BackgroundArthroscopic release for elbow stiffness is considered a minimally invasive and effective treatment. However, the extent to which each intraoperative step contributes to improvement in range of motion (ROM) has not been well investigated. PurposeTo sequentially evaluate the relationship between intraoperative surgical steps and changes in elbow ROM during arthroscopic release for severe elbow stiffness, and to identify the key procedural stage contributing most significantly to ROM improvement. MethodsFive elbows in five patients with severe elbow stiffness following fracture or dislocation were retrospectively reviewed. Arthroscopic release was performed using a stepwise posterior-based approach, starting from the posterior soft-spot portal, followed by exposure of the olecranon fossa and progression into the posteromedial compartment. Changes in elbow ROM were assessed at each intraoperative step, and ROM at final follow-up was also evaluated. ResultsAll patients demonstrated improvement in elbow ROM at final follow-up. Intraoperative ROM improvement did not occur in a continuous manner but rather in a stepwise fashion. Gradual improvement was observed with establishment of the posterior and posteromedial working spaces, followed by the most substantial increase in ROM immediately after release of the soft tissue attached to the posterior aspect of the humeral medial epicondyle. Although the maximum ROM achieved intraoperatively was not fully maintained at final follow-up, no patient experienced deterioration to preoperative ROM levels. ConclusionsIn arthroscopic release for severe elbow stiffness, improvement in elbow ROM occurs in a stepwise rather than continuous pattern. Release of the posteromedial structures attached to the posterior aspect of the humeral medial epicondyle may represent a critical turning point contributing significantly to ROM improvement.

A critical challenge in endocrine neurosurgery is intraoperative discrimination between normal pituitary tissue and pituitary neuroendocrine tumors (PitNETs). Suggesting the universal persistence of near-infrared autofluorescence (NIRAF) in endocrine organs and inspired by routine clinical use of NIRAF for parathyroid gland identification, we discovered that pituitary NIRAF can be employed for label-free transsphenoidal surgery guidance. Ex vivo confocal spectral imaging of 33 specimens identified secretory granules as the dominant long-wavelength fluorescence source and showed that normal pituitary had higher granule content than PitNETs. For the first time, we made use of the pituitary NIRAF during surgery and assessed its performance for pituitary/adenoma separation in vivo for 27 surgeries and showed near-perfect separability between pituitary and non-pituitary measurement sites with ROC-AUC of 0.98. The obtained results clearly demonstrate that the suggested method, based on the solid microscopic background, has the potential for clinical translation and paves the way for enhanced gland preservation during resection.

The use of semaglutide (SE), a glucagon-like peptide-1 receptor agonist (GLP-1RA) with glucose-lowering and weight-loss effects, has risen rapidly, particularly among women of reproductive age. While preclinical studies suggest benefits for ovarian function via the hypothalamic-pituitary-ovarian axis, its impact on the endometrial-embryo interface remains unclear. Here, we show that GLP-1R is dynamically expressed in fertile human endometrium, restricted to epithelial cells and markedly upregulated during the mid-secretory phase of the menstrual cycle. In a preclinical model of endometrial epithelial organoids, SE at physiological concentrations activates intracellular cAMP signaling, enhances epithelial metabolism, and upregulates receptivity markers without steroid hormone priming, whereas higher concentrations modestly reduce expression of a key receptivity marker PAEP/glycodelin and shift metabolism towards oxidative phosphorylation. By contrast, in stromal cells lacking detectable GLP-1R, SE disrupts decidualization, induces endoplasmic reticulum stress and suppresses cell-cycle at G2/M phase. Human embryo models, blastoids, expressed GLP-1R and underwent concordant SE-mediated transcriptional remodeling in epiblast and trophectoderm lineages, encompassing changes in metabolism and epigenetic regulation, but without shifts in lineage proportions. Notably, SE increased blastoid attachment to the endometrial epithelium in the absence of exogenous steroid hormones, suggesting enhanced epithelial-embryo interaction. Together, these findings reveal a compartment-specific mismatch, as SE augments epithelial and embryonic metabolic activity but compromises stromal support for implantation, with potential consequences for implantation due to stromal dysfunction.

BackgroundPlacental growth and function are imperative for healthy fetal growth; data on placentas can inform research and clinical care. Measuring placental size after delivery should be easy, but current methods are hard to standardize and error prone. We developed PlacentaVision using artificial intelligence (AI)-based models, to automatically, accurately, and precisely measure placentas from digital photographs. ObjectiveWe aimed to compare placental disc morphology between gross pathology examination (human measurements) and our automated PlacentaVision model (AI measurements). MethodsPlacentaVision is a multi-site study to assess placental morphology, features, and pathologies from digital photographs. We built a large dataset of digital placenta photographs and clinical data from singleton births at three large hospitals: Northwestern Memorial (Chicago; n=24,933), UPMC Magee-Womens (Pittsburgh; n=1198) and Mbarara Regional Referral (Uganda, n=1715). Data and images were from the medical record for Northwestern, part of a biobank study for Magee, and from our prospective studies for Mbarara. We compared long and short disc axis length (defined by Amsterdam criteria) between human and AI-based PlacentaVision measurements by calculating the difference and using Bland-Altman; we stratified by site, disc shape, infant sex, and term/preterm birth. ResultsMean (SD) disc length was 19.2 (3.1) and 18.6 (3.1) cm from PlacentaVision and human measurement, respectively, with a difference of 0.57 (2.19) cm. Disc width was 16.3 (2.3) cm and 16.1 (2.4) cm from PlacentaVision and human measurement, respectively, with a difference of 0.25 (1.85) cm. Bland-Altman limits of agreement were -3.7 to 4.9 cm for length and -3.4 to 3.9 cm for width. Irregularly-shaped placentas had a greater difference between PlacentaVision and human measurements compared to those with round/oval shapes (length differences of 1.53 and 0.45 cm respectively). Further, there were length differences by site (Northwestern 0.6, Magee 0.0, and Mbarara 0.4) and gestational age at birth (preterm 0.71, term 0.53 cm), but similar results for male and female placentas. Results for width were similar to length. ConclusionsAI-based measurements were less than a cm from human measurements overall. Our findings of larger differences for irregular shapes and preterm may indicate it is difficult for humans to measure irregular or small placentas according to protocol. PlacentaVision can automate and standardize the process.

IntroductionAdverse drug reactions (ADRs) remain a major public health issue, and genetic factors contribute importantly to interindividual variability in drug response. Pharmacogenetic testing helps reduce ADR risk by optimizing drug selection and dosage, particularly in monogenic disorders. Material and MethodsWhole-exome sequencing of 6,739 samples from the Russian population was performed using the MGIEasy Universal DNA Library Prep Set on the DNBSEQ-G400 platform (MGI). Variants in 48 genes were examined, focusing on inherited arrhythmias (Long QT syndrome, Short QT syndrome, Timothy syndrome, Andersen-Tawil syndrome, Brugada syndrome, Atrial fibrillation, Catecholaminergic polymorphic ventricular tachycardia), enzyme deficiencies (Glucose-6-Phosphate Dehydrogenase Deficiency [G6PDD], Porphyrias), Dravet Syndrome (DS) and Malignant Hyperthermia (MH). All identified variants had been reported at least once as pathogenic (P) or likely pathogenic (LP) in ClinVar, along with those occasionally classified as variants of uncertain significance (VUS). Each variant was manually re-evaluated according to ACMG criteria. ResultsA total of 75 unique variants in 18 genes were observed in 119 individuals (1.77%), including 21 carriers and 13 women with a G6PD mutation. Of these, 46 variants were classified as P, 21 as LP, and 8 as VUS. Missense variants accounted for the largest proportion (73.33%). The most affected genes were KCNQ1 (24/119), which exhibited the highest number of unique variants (18), G6PD (20/119), SCN1A (15/119), and RYR1 (14/119). Regarding associated conditions, mutations linked to arrhythmias were found in 51 individuals, MH in 27, G6PDD in 20, DS in 15, and Porphyrias in 6. ConclusionsIncorporating genetic information on both common and rare clinically actionable variants into therapeutic decision-making has the potential to improve medication safety, reduce preventable ADRs, and enhance the effectiveness of personalized pharmacotherapy.

Topiramate (TPM) is approved for seizures and migraine prophylaxis and is used off-label for several neuropsychiatric conditions. The available dosage forms, including tablets and sprinkle capsules, are unsuitable for patients who may be unable to take medicine orally. The resulting potential treatment interruptions could have untoward consequences and underscores the importance of developing a parenteral formulation. In this study, we developed a population pharmacokinetic model of a novel, intravenous TPM formulation using data from a study in patients with epilepsy or migraine receiving a single intravenous dose of stable-labeled TPM. In total, 246 TPM concentrations from 20 adult patients were included for model development. A three-compartment pharmacokinetic model with linear elimination fit the concentration-time data best. Simulations for various loading and maintenance regimens for patients with and without enzyme-inducing comedications were performed. The final estimates(95% confidence interval (CI)) for CL (L/h), V1 (L), and the peripheral volumes, V2 and V3 for a 70 kg person were 1.31(1.01 - 1.53), 9.84 (8.49 - 11.0), 39.1 (36.5 - 41.8)L, and 9.01 (6.41 - 44.3) respectively. The use of enzyme-inducing co-medication was the only significant covariate, associated with a 63% increase in clearance .Goodness-of-fit plots and visual predictive checks indicate satisfactory model performance and prediction. The simulation results indicate that adjusting doses for patients receiving IV TPM can mitigate the changes in plasma TPM concentrations resulting from enzyme induction. This population pharmacokinetic model for intravenous topiramate can inform dosing decisions for patients with epilepsy when used as either initiation or bridging therapy.

BackgroundPersons with epilepsy are at increased risk of depression/anxiety. Older antiseizure medications (ASMs) had drug-drug interactions that complicated pharmacotherapy of depression/anxiety; newer ASMs lack this drawback but can have psychiatric side effects. Anxiety/depression are increasingly recognized and treated pharmacologically. We hypothesized that the likelihood of treatment with selective serotonin uptake inhibitors (SSRI) would have increased in adult-onset epilepsy when prescription habits shifted towards newer ASMs. MethodsWe linked national health registers and included 28569 persons with epilepsy incident in 2006-2020 and 68509 age- and sex matched controls. We assessed the risk of starting SSRI treatment compared to age- and sex-matched controls across three incidence periods: 2006-2010, 2011-2015, and 2016-2020. Cox regression was used to estimate adjusted hazard ratios (HRs), and subgroup analyses explored age, sex, and comorbidities. Specialist psychiatric care was also assessed as a measure of more severe depression. Analysis including persons with SSRI-use before the epilepsy diagnosis were used for sensitivity analyses. FindingsPersons with epilepsy had higher risks of starting SSRIs compared to controls; 1986/9561 (20.8%) received SSRI during follow-up after epilepsy in 2006-2010 and 2020/9165 (22.0%) in 2016-2020; adjusted HRs were 1.92 (95%CI:1.79 - 2.06) in 2006-2010, 1.84 (95%CI:1.72-1.97) in 2011-2015, and 1.81 (95%CI:1.69 - 1.94) in 2016-2020. Among individuals aged 18-30 years at their epilepsy diagnosis, the proportion receiving SSRIs remained the same between the first and last calendar periods (18.2%). Because of increased treatment of controls, the adjusted HRs of SSRI-treatment decreased from 2.33, (95% CI:1.96 - 2.78) to 1.63, (95% CI 1.39 to 1.91). The HR of specialist psychiatric care was not significantly different between the time periods. Most comorbidities were consistently associated with increased likelihood of SSRI treatment, whereas intellectual disability decreased the likelihood in some periods. InterpretationWe found no evidence of overall increased SSRI initiation or psychiatric care after the shift to newer ASMs. Person with epilepsy remain more likely to receive SSRI treatment, but probably not to a level matching the higher prevalence of depression. Increased SSRI treatment of younger age adults has not been matched by increased treatment of young adults with epilepsy. This suggests a potentially widening treatment gap and a need for increased recognition of depression in young adults with epilepsy. FundingSwedish Research Council (2023-02816), Swedish state through the ALF-agreement (ALFGBG-1006343), Knut och Ragnvi Jacobsson foundation, Swedish Society for Medical Research (S18-0040), Swedish Society of medicine (SLS-881501), Epilepsifonden, Rune och Ulla Amlovs stiftelse.

Applying deep learning models to RNA-Seq data poses substantial challenges, primarily due to the high dimensionality of the data and the limited sample sizes. To address these issues, this study introduces an advanced deep learning pipeline that integrates feature engineering with data augmentation. The engineering application focuses on biomedical engineering, specifically the classification of RNA-Seq datasets for disease diagnosis. The proposed framework was initially validated on synthetic datasets generated from Naive Bayes, where MLP-based augmentation yielded a notable improvement in predictive performance. Building on this foundation, we applied the approach to chromophobe renal cell carcinoma (KICH) RNA-Seq data from The Cancer Genome Atlas (TCGA). Following standard preprocessing steps normalization, transformation, and dimensionality reduction, the analysis concentrated on three main aspects: augmentation strategies, preprocessing methods, and explainable AI (XAI) techniques in relation to classification outcomes. Feature selection was performed through PCA, Boruta, and RF-based methods. Three augmentation strategies linear interpolation, SMOTE, and MixUp were evaluated. To maintain methodological rigor, augmentation was applied exclusively to the training set, while the test set was held out for unbiased evaluation. Within this framework, we conducted a comparative assessment of multiple deep learning architectures, including MLP, GNN, and the recently proposed Kolmogorov-Arnold networks (KAN). The GNN achieved the highest classification accuracy (99.47%) when trained with MixUp augmentation combined with RF feature selection, and achieved the best F1 score (0.9948). Consequently, the GNN-based XAI framework was applied to the RF dataset enriched with MixUp. XAI analyses identified the top 20 most influential genes, such as HNF4A, DACH2, MAPK15, and NAT2, which played the greatest role in classification, thereby confirming the biological plausibility of the model outputs. To further validate model robustness, cervical cancer and Alzheimers RNA-Seq datasets were also tested, yielding consistent and reliable results. Overall, the findings highlight the value of incorporating data augmentation into deep learning models for RNA-Seq analysis, not only to improve predictive performance but also to enhance biological interpretability through explainable AI approaches.

AO_SCPLOWBSTRACTC_SCPLOWO_ST_ABSINTRODUCTIONC_ST_ABSTreatment response in Alzheimers disease (AD) varies substantially across patients, yet no validated frameworks exist to estimate heterogeneous treatment effects (HTE) from observational data while controlling for confounding bias. METHODSWe developed a causal machine learning framework integrating expert-guided causal graphs, complementary HTE estimators, sensitivity analyses, and policy learning. We applied it to cholinesterase inhibitors (ChEIs) in MCI due to AD to patients from the NACC and ADNI cohorts. RESULTSAnalysing 4,049 patients with 12-month and 2,223 with 36-month follow-up, all estimators indicated null or negative long-term ChEI effects on cognitive and functional outcomes, notably on functional measures. ChEIs showed slightly more deleterious effects among men than women. DISCUSSIONThis framework provides a methodology for estimating HTE from observational data. It revealed no beneficial responder subgroups, highlighting the challenge of detecting treatment heterogeneity in moderately sized cohorts. This approach can inform treatment selection for other AD therapies including memantine, anti-amyloid agents, and emerging treatments.

Meningiomas are the most common primary brain tumors and, despite their benign reputation, often behave aggressively. Meningiomas are morphologically heterogeneous, yet the full significance of their histologic diversity is unclear. This is in large part because many features are not readily quantifiable by traditional observer-based light microscopy. Molecular testing improves prognostic stratification, but is not universally accessible. We therefore sought to determine whether an artificial intelligence (AI)-trained program could predict specific genomic and epigenomic patterns in meningiomas, and whether it could extract more prognostic information out of standard hematoxylin and eosin (H&E) histopathology than the current WHO classification. To do this, we developed Morphologic Set Enrichment (MSE), an interpretable computational pathology framework that quantifies statistical enrichment of morphologic patterns, cells, and tissue architecture from H&E whole-slide images. The MSE meningioma histology program was able to accurately predict DNA methylation subtypes and concurrent chromosome 1p/22q losses, in the process identifying specific morphologic patterns associated with key genomic and epigenomic alterations. It also added prognostic value independent of standard clinical and pathological variables. These results demonstrate that AI-based quantitative morphologic profiling can capture clinically and biologically relevant information that redefines risk stratification for meningiomas, incorporating histological information not included in existing grading schemes.

ABSTRACT/SUMMARYO_ST_ABSObjectiveC_ST_ABSCannabis use during pregnancy is increasing; associations with neonatal growth may be confounded by nicotine. We evaluated prenatal cannabis exposure (PreCE) and neonatal outcomes in a prospective cohort with biochemical control for nicotine exposure. MethodsIn the Cannabis Use During Early Life and Development (CUDDEL) study, pregnant women with a lifetime history of cannabis use were classified as PreCE if they self-reported use or had urine THC-COOH positivity at any trimester (n=297) and as unexposed if they reported no use and tested negative (n=151). Linear regression and modified Poisson models estimated associations with birthweight and small for gestational age (SGA; <10th and <5th percentiles), adjusting for sociodemographic factors, gestational age, maternal age and BMI, and urinary cotinine. Analyses stratified by cannabis use frequency (>weekly vs <monthly) and cotinine status. ResultsParticipants (N=448; 18-41 years; 85.3% non-Hispanic Black) had lower birthweight with PreCE in adjusted models (Beta=-0.08; padj=0.041). High-frequency PreCE was associated with lower birthweight compared with unexposed pregnancies (Beta=-0.13; padj=0.03), whereas low-frequency PreCE was not. Cotinine-positive PreCE showed the greatest birthweight reduction versus unexposed (Beta=-0.20; padj<0.001). PreCE was also associated with higher likelihood of SGA <5th percentile; risk was highest in PreCE+Nicotine compared with both unexposed and PreCE-Nicotine groups. ConclusionsPrenatal cannabis exposure was associated with reduced birthweight and SGA in this cohort. Nicotine co-exposure intensified these associations, yet effects persisted without cotinine, supporting cannabis as an independent perinatal risk factor and emphasizing the value of cotinine assessment in populations where blunt use or secondhand exposure is common.

BackgroundThe Antidepressant Medications: Biology, Exposure & Response (AMBER) research programme was established to investigate the biological mechanisms underlying antidepressant action and variability in treatment response. Generation Scotland holds detailed genomic, clinical, and health information with recontacting consent, making this cohort ideal for investigating these aims. MethodsWe deployed a questionnaire, developed with input from a Lived Experience panel, to the Generation Scotland cohort to gather data on their depressive symptoms, medication history, efficacy, and side effects to develop clinically meaningful phenotypes of antidepressant response. Invitations were sent to 15,117 Generation Scotland participants who were 18 years or older and consented to be recontacted. Between July and November 2025, 1,180 participants with a history of antidepressant treatment for depression completed the questionnaire. ResultsThe sample was predominantly female (78.1%), self-identified as White (98.6%), and older (median age 57 years) than the wider Generation Scotland cohort (median 49 years) and Scottish population (median 41.3 years). Participants reported heterogeneous depressive symptom profiles spanning mood, anxiety, cognitive, sleep, behavioural, and physical domains. One-third of participants (31.1%) had taken three or more different antidepressants. Selective serotonin reuptake inhibitors (SSRIs) were the most common class (89.1%). Using self-reported treatment duration, discontinuation patterns, and efficacy, we developed a stringent classification system to capture treatment response extremes, where 23.8% were classified as responders and 1.5% as non-responders, with the majority unclassified. ConclusionsQuestionnaire data will be linked with electronic health records to validate antidepressant response classifications. Following validation, 25 responders and 25 non-responders will provide biological samples for DNA methylation profiling and generation of patient-derived cell lines. These models will be exposed to SSRIs to identify gene expression signatures and biological pathways distinguishing treatment response, integrating with genomic and clinical data across the AMBER project. These findings will provide a valuable resource for future antidepressant response research. Plain Language SummaryDepression is a common mental health condition affecting millions of people worldwide. Antidepressant medications are the primary medication treatment, but response is highly variable with only about one-third of individuals achieving full symptom remission after their first medication trial. We dont fully understand why some people respond well while others dont. To help answer this question, the Antidepressant Medications: Biology, Exposure & Response (AMBER) research programme was established. This study utilised the Generation Scotland cohort, a large health study in Scotland. Between July and November 2025, we invited 15,117 Generation Scotland participants to complete a detailed questionnaire about their experiences with antidepressant medications. A total of 1,180 participants answered detailed questions about their depression symptoms, which medications they tried, how long they were on a medication, how well the medications worked, and what side effects they experienced. We found that peoples experiences with depression and antidepressants varied considerably. About one-third had tried three or more different antidepressants. Using strict criteria based on treatment duration, effectiveness ratings, and medication changes, we identified 281 people (24%) who responded very well to SSRIs (the most common type of antidepressant) and 18 people (1.5%) who did not respond despite trying multiple SSRIs. A key limitation is that all information was self-reported, so we will validate findings by linking questionnaire responses with medical records. In the future, we will collect blood samples from some participants to study the biological differences between responders and non-responders. This research will help us better understand why antidepressants work for some people but not others, which could lead to more personalised treatment approaches for depression.

Chemical shift-encoded magnetic resonance imaging using high-resolved 3D Dixon techniques enables the non-invasive and radiation-free assessment of whole-body adipose tissue and ectopic fat distribution. Automatic deep learning-based segmentation of metabolically relevant adipose tissue compartments and ectopic fat deposits in parenchymal tissue is the most important image processing step for the quantification of adipose tissue volumes and ectopic fat percentages from whole-body imaging. This work presents a segmentation model dedicated to the segmentation of 19 metabolically relevant adipose tissue compartments and ectopic fat deposits from whole-body Dixon MRI. The trained segmentation model is available upon request. Related post-processing routines to compute volumes and fat percentages are publicly available: https://github.com/tobihaui/WholeBodyATQuantification.