Sci Rep. 2025 Apr 18;15(1):13502. doi: 10.1038/s41598-025-97113-0.

ABSTRACT

Heart failure (HF) remains a leading global cause of cardiovascular deaths, with its prevalence expected to rise in the upcoming decade. Measuring the heart ejection fraction (EF) is crucial for diagnosing and monitoring HF. Although echocardiography is the gold standard for EF measurement, it is often inaccessible in remote areas due to its cost and complexity. In contrast, electrocardiography (ECG) is more readily available and affordable, and emerging research suggests a possible link between ECG signals and EF. In this work, we explore the potential of 12-lead ECG signals to estimate EF using various machine learning (ML) and deep learning (DL) models. While recent studies have considered the use of ML or DL for estimating EF, these algorithms are often trained and tested on urban-based populations. However, demographics like those in rural Appalachia, where disease prevalence is extremely high, have been overlooked, potentially due to the unavailability of large volumes of data. Moreover, there have been concerning reports regarding the fairness of AI predictions across different populations, making it crucial to understand the performance of AI models across diverse demographics before their widespread application. To address this, our study focuses on analyzing AI models for EF estimation in the rural Appalachian population. We utilized a 12-lead ECG dataset of 55,500 patients from WVU Medicine hospitals in West Virginia and employed a wide array of AI algorithms, ranging from Random Forest to modern deep learning-based methods like Transformers, to estimate EF. We also considered different thresholds for analyzing these AI algorithms and examined the impact of single and multi-lead ECG signals, and conducted model interpretability analysis. Overall, our comprehensive analysis demonstrated that deep learning-based algorithms achieved the highest performance, with an AUROC of around 0.86 for EF estimation from 12-lead ECG signals. Additionally, we found that while individual ECG leads were insufficient for accurate EF estimation, specific lead combinations significantly improved classification performance.

PMID:40251349 | DOI:10.1038/s41598-025-97113-0

Sci Rep. 2025 Apr 18;15(1):13429. doi: 10.1038/s41598-025-97885-5.

ABSTRACT

Self-reported pain scores are often used for pain assessments and require effective communication. On the other hand, observer-based assessments are resource-intensive and require training. We developed an automated system to assess pain intensity in adult patients based on changes in facial expression. We recruited adult patients undergoing surgery or interventional pain procedures in two public healthcare institutions in Singapore. The patients' facial expressions were videotaped from a frontal view with varying body poses using a customized mobile application. The collected videos were trimmed into multiple 1 s clips and categorized into three levels of pain: no pain, mild pain, or significant pain. A total of 468 facial key points were extracted from each video frame. A customized spatial temporal attention long short-term memory (STA-LSTM) deep learning network was trained and validated using the extracted keypoints to detect pain levels by analyzing facial expressions in both the spatial and temporal domains. Model performance was evaluated using accuracy, sensitivity, recall, and F1-score. Two hundred patients were recruited, with 2008 videos collected for further clipping into 10,274 1 s clips. Videos from 160 patients (7599 clips) were used for STA-LSTM training, while the remaining 40 patients' videos (2675 clips) were set aside for validation. By differentiating the polychromous levels of pain (no pain versus mild pain versus significant pain requiring clinical intervention), we reported the optimal performance of STA-LSTM model, with accuracy, sensitivity, recall, and F1-score all at 0.8660. Our proposed solution has the potential to facilitate objective pain assessment in clinical settings through the developed STA-LSTM model, enabling healthcare professionals and caregivers to perform pain assessments effectively in both inpatient and outpatient settings.

PMID:40251301 | DOI:10.1038/s41598-025-97885-5

Sci Rep. 2025 Apr 18;15(1):13428. doi: 10.1038/s41598-025-98021-z.

ABSTRACT

While deep learning has become the go-to method for image denoising due to its impressive noise removal Retinal blood vessel segmentation presents several challenges, including limited labeled image data, complex multi-scale vessel structures, and susceptibility to interference from lesion areas. To confront these challenges, this work offers a novel technique that integrates attention mechanisms and a cascaded dilated convolution module (CDCM) within a multi-path U-Net architecture. First, a dual-path U-Net is developed to extract both coarse and fine-grained vessel structures through separate texture and structural branches. A CDCM is integrated to gather multi-scale vessel features, enhancing the model's ability to extract deep semantic features. Second, a boosting algorithm that incorporates probability distribution attention (PDA) within the upscaling blocks is employed. This approach adjusts the probability distribution, increasing the contribution of shallow information, thereby enhancing segmentation performance in complex backgrounds and reducing the risk of overfitting. Finally, the output from the dual-path U-Net is processed through a feature refinement module. This step further refines the vessel segmentation by integrating and extracting relevant features. Results from experiments on three benchmark datasets, including CHASEDB1, DRIVE, and STARE, demonstrate that the proposed method delivers improved segmentation accuracy compared to existing techniques.

PMID:40251298 | DOI:10.1038/s41598-025-98021-z

Sci Rep. 2025 Apr 18;15(1):13445. doi: 10.1038/s41598-025-98629-1.

ABSTRACT

Traditional drug design faces significant challenges due to inherent chemical and biological complexities, often resulting in high failure rates in clinical trials. Deep learning advancements, particularly generative models, offer potential solutions to these challenges. One promising algorithm is DrugGPT, a transformer-based model, that generates small molecules for input protein sequences. Although promising, it generates both chemically valid and invalid structures and does not incorporate the features of approved drugs, resulting in time-consuming and inefficient drug discovery. To address these issues, we introduce DrugGen, an enhanced model based on the DrugGPT structure. DrugGen is fine-tuned on approved drug-target interactions and optimized with proximal policy optimization. By giving reward feedback from protein-ligand binding affinity prediction using pre-trained transformers (PLAPT) and a customized invalid structure assessor, DrugGen significantly improves performance. Evaluation across multiple targets demonstrated that DrugGen achieves 100% valid structure generation compared to 95.5% with DrugGPT and produced molecules with higher predicted binding affinities (7.22 [6.30-8.07]) compared to DrugGPT (5.81 [4.97-6.63]) while maintaining diversity and novelty. Docking simulations further validate its ability to generate molecules targeting binding sites effectively. For example, in the case of fatty acid-binding protein 5 (FABP5), DrugGen generated molecules with superior docking scores (FABP5/11, -9.537 and FABP5/5, -8.399) compared to the reference molecule (Palmitic acid, -6.177). Beyond lead compound generation, DrugGen also shows potential for drug repositioning and creating novel pharmacophores for existing targets. By producing high-quality small molecules, DrugGen provides a high-performance medium for advancing pharmaceutical research and drug discovery.

PMID:40251288 | DOI:10.1038/s41598-025-98629-1

Sci Rep. 2025 Apr 18;15(1):13501. doi: 10.1038/s41598-025-98523-w.

ABSTRACT

Breast cancer is a leading killer and has been deepened by COVID-19, which affected diagnosis and treatment services. The absence of a rapid, efficient, accurate diagnostic tool remains a pressing issue for this severe disease. Thus, it is still possible to encounter issues concerning diagnostic accuracy and utilization of errors in the sphere of machine learning, deep learning, and transfer learning models. This paper presents a new model combining EfficientNetB0 and ResNet50 to improve the classification of breast histopathology images into IDC and non-IDC classes. The implementation steps, it include resizing all the images to be of a standard size of 128*128 pixels and then performing normalization to enhance the learning model. EfficientNetB0 is selected for its efficient yet effective performance while ResNet50 employs deep residual connections to overcome the vanishing gradient problem. The proposed model that incorporates some of the characteristics from both architectures turns out to be very resilient and accurate in classification. The model demonstrates superior performance with an accuracy of 94%, a Mean Absolute Error (MAE) of 0.0628, and a Matthews Correlation Coefficient (MCC) of 0.8690. These results outperform previous baselines and show that the model performs well in achieving a good trade-off between precision and recall. The comparison with the related works demonstrates the superiority of the proposed ensemble approach in terms of accuracy and complexity, which makes it efficient for practical breast cancer diagnosis and screening.

PMID:40251247 | DOI:10.1038/s41598-025-98523-w

Respir Investig. 2025 Apr 17;63(4):481-487. doi: 10.1016/j.resinv.2025.04.008. Online ahead of print.

ABSTRACT

BACKGROUND: The Japanese government operates a medical subsidy system for intractable diseases, including idiopathic pulmonary fibrosis (IPF). Registering with this system requires filling out a clinical survey form, which encompasses multiple survey items regarding the patient's disease and functional status. In this study, we retrospectively analyzed the forms of new applicants with IPF in 2019 to evaluate the clinical and functional characteristics.

METHODS: The following patient data were collected: sex, age, smoking status, history of surgical lung biopsy, disease severity (using Japanese severity classification [JSC]), serum biomarkers, findings on chest high-resolution computed tomography (HRCT), functional status, symptoms, and treatment.

RESULTS: Of the 4796 patients reviewed (76.1 % males; mean age = 73.4 ± 8.1 years), 23.6 % had a mild disease (JSC stages I-II) and 76.4 % had a severe disease (stages III-IV). The HRCT of most patients revealed honeycombing, traction bronchiectasis and/or bronchiolectasis, reticular shadows, and subpleural shadows. The positivity rates for elevated serum levels of Krebs von Lungen-6 (KL-6) and surfactant protein-D (SP-D) were 92.6 % and 89.3 %, respectively. As the severity increased, the biomarker positivity rate increased. Approximately half of the patients with milder diseases experienced transportation challenges, and 30 % complained of pain and/or discomfort and anxiety and/or depression.

CONCLUSIONS: In approximately 90 % of patients, serum KL-6 and SP-D levels increased and the positive rate increased as the disease severity increased. Even patients with mild diseases experience challenges in transportation as well as pain, discomfort, anxiety, or depression.

PMID:40250140 | DOI:10.1016/j.resinv.2025.04.008

Sci Rep. 2025 Apr 18;15(1):13448. doi: 10.1038/s41598-025-96687-z.

ABSTRACT

JQ1, a small cell-permeable molecule is known for its potent inhibitory action on bromodomain and extraterminal (BET) proteins. Although earlier studies have shown its inhibitory effect on male gametogenesis, limited information is available about its influence on oocyte development. Since BET genes are known to exhibit regulatory functions on oocyte development and maturation, the present study aimed to investigate the effect of JQ1 on oocyte developmental competence under in vitro conditions. Germinal vesicle (GV) stage oocytes were collected from adult Swiss albino mice and subjected to in vitro maturation (IVM) in the presence of various concentrations of JQ1 (25, 50, and 100 μM). The metaphase II (MII) stage oocytes were assessed for cytoplasmic organization and functional competence at 24 h after IVM. A significant decrease in nuclear maturation (at 50 and 100 μM), symmetric cytokinesis, altered distribution of mitochondria and cortical granules, poorly organized actin and meiotic spindle, misaligned chromosomes, and elevated endoplasmic reticulum (ER) stress and oxidative stress was observed in JQ1-exposed oocytes. Presence of N-acetyl cysteine (NAC), in IVM medium resulted in significant reduction in JQ1-induced oxidative stress and symmetric cytokinesis. Administration of JQ1 (50 mg/kg, intra peritoneal) to adult Swiss albino mice primed with pregnant mare serum gonadotrophin (PMSG) and human chorionic gonadotrophin (hCG) did not affect the ovulation. However, a high degree of oocyte degeneration, elevated intracellular reactive oxygen species (ROS), and GRP78 expression was observed in JQ1-administered mice. In conclusion, our study reveals that BET inhibitor JQ1 has detrimental effects on oocyte function and development.

PMID:40251236 | DOI:10.1038/s41598-025-96687-z

Nat Commun. 2025 Apr 18;16(1):3700. doi: 10.1038/s41467-025-58878-0.

ABSTRACT

How cells change shape is crucial for the development of tissues, organs and embryos. However, studying these shape changes in detail is challenging. Here we present a comprehensive real-time cellular map that covers over 95% of the cells formed during Caenorhabditis elegans embryogenesis, featuring nearly 400,000 3D cell regions. This map includes information on each cell's identity, lineage, fate, shape, volume, surface area, contact area, and gene expression profiles, all accessible through our user-friendly software and website. Our map allows for detailed analysis of key developmental processes, including dorsal intercalation, intestinal formation, and muscle assembly. We show how Notch and Wnt signaling pathways, along with mechanical forces from cell interactions, regulate cell fate decisions and size asymmetries. Our findings suggest that repeated Notch signaling drives size disparities in the large excretory cell, which functions like a kidney. This work sets the stage for in-depth studies of the mechanisms controlling cell fate differentiation and morphogenesis.

PMID:40251161 | DOI:10.1038/s41467-025-58878-0

Trends Biochem Sci. 2025 Apr 17:S0968-0004(25)00052-0. doi: 10.1016/j.tibs.2025.03.003. Online ahead of print.

ABSTRACT

Brassinosteroids (BRs) are essential plant steroidal hormones that regulate growth and development. The recent discoveries of ATP-binding cassette subfamily B (ABCB) members, ABCB19 and ABCB1, as BR transporters highlight the significance of active export to the apoplast in maintaining BR homeostasis and enabling effective signaling. This review focuses on the latest progress in understanding ABCB-mediated BR transport, with particular attention to the structural and functional characterization of arabidopsis ABCB19 and ABCB1. These findings reveal both conserved and distinct features in substrate recognition and transport mechanisms, providing valuable insights into their roles in hormonal regulation. Additionally, the evolutionary conservation of ABC transporters in mediating steroid-based signaling across biological kingdoms underscores their fundamental biological significance.

PMID:40251078 | DOI:10.1016/j.tibs.2025.03.003

Food Chem Toxicol. 2025 Apr 16:115457. doi: 10.1016/j.fct.2025.115457. Online ahead of print.

ABSTRACT

The liver is considered a target organ for the accumulation and toxic effects of nanomaterials exposed to the body, especially after oral exposure, but the key toxic pathways have not been fully defined. This study focused on the hepatotoxicity of titanium dioxide nanoparticles (TiO2 NPs) in vivo and in vitro, and tried to identify key toxic pathways using the concept of systems biology and multi-omics methods. In vivo, protein and metabolomic sequencing were performed on the liver of SD rats (0, 50 mg/kg, 90 days), and 386 differential proteins and 29 differential metabolites were screened out, respectively, and the joint analysis found that they were significantly enriched in alanine, aspartate and glutamate metabolism, and butanoate metabolism. In vitro, exposure to TiO2 NPs could induce cytotoxicity and omics changes at different molecular levels in human hepatocellular carcinoma cells. Single omic analysis showed that differentially expressed proteins and metabolites were 80 and 222, respectively. The enriched pathways related to steroid biosynthesis, cholesterol metabolism at the combine levels of proteome and metabolome. KEGG enrichment analysis showed that PI3K-Akt signaling pathway and PPAR signaling pathway were both significantly affected in vitro and in vivo. Through multi-omics analysis, this work offered fresh perspectives and avenues for research on the toxicity mechanism of TiO2 NPs.

PMID:40250523 | DOI:10.1016/j.fct.2025.115457

Mol Cell. 2025 Apr 17;85(8):1477-1479. doi: 10.1016/j.molcel.2025.03.022.

ABSTRACT

In this issue of Molecular Cell, Schärfen et al.1 describe an advanced RNA structure-probing technology called CoSTseq that enables transcriptome-wide detection of nascent RNA base pairing during transcription in living yeast cells.

PMID:40250407 | DOI:10.1016/j.molcel.2025.03.022

Int Wound J. 2025 Apr;22(4):e70513. doi: 10.1111/iwj.70513.

ABSTRACT

Tyrosine kinase inhibitors (TKIs) can treat various cancers, primarily through their antiangiogenic effects. However, as angiogenesis is crucial for successful wound healing, TKIs may adversely impact wound repair. This review analysed all 63 FDA-approved TKIs and identified evidence for wound healing and repair implications in 24 agents. The primary mechanism contributing to impaired wound healing appears to be the inhibition of vascular endothelial growth factor receptors, with secondary targets, such as epidermal growth factor receptors and platelet-derived growth factor receptors, potentially playing a role. Information from safety package inserts, preclinical studies, case reports and clinical trials suggests that these TKIs can cause delayed or impaired wound healing. The safety information generally recommends discontinuing treatment for at least one to 2 weeks before elective surgery and resuming treatment only after adequate wound healing has occurred. Neoadjuvant therapy with TKIs may be feasible if sufficient time is allowed between the cessation of the TKI and the onset of surgery. As the use of TKIs continues to increase, healthcare professionals should be aware of their potential impact on wound healing and take appropriate precautions to minimise the risk of wound-related complications.

PMID:40251464 | DOI:10.1111/iwj.70513

Sci Rep. 2025 Apr 18;15(1):13407. doi: 10.1038/s41598-025-92796-x.

ABSTRACT

Drug-related conjunctivitis can compromise ocular health and quality of life. To evaluate its epidemiology, we analyzed reports from the FDA Adverse Event Reporting System (FAERS) spanning January 2004 to June 2024. The control group in this study comprised individuals using non-target drugs, while the experimental group consisted of individuals using target drugs. Using disproportionality analysis, we identified drugs with a positive signal for conjunctivitis and stratified their risk levels; we also examined induction periods to assess the speed of onset. Among 38 drugs most frequently reported for conjunctivitis, two ophthalmic agents-brimonidine (ROR = 23.04) and latanoprost (ROR = 10.55)-and eight non-ophthalmic drugs, including tralokinumab (ROR = 83.3), dupilumab (ROR = 18.92), and allopurinol (ROR = 5.04), were associated with positive signals. Tralokinumab, brimonidine, dupilumab, and latanoprost were identified as high-association medications. Notably, ophthalmic agents had a significantly shorter induction period than non-ophthalmic drugs (mean 125.9 vs. 298.4 days). These findings underscore the need for vigilant pharmacovigilance and further investigation into the etiology and prevention of drug-related conjunctivitis.

PMID:40251175 | DOI:10.1038/s41598-025-92796-x

Eur J Hosp Pharm. 2025 Apr 18:ejhpharm-2025-004475. doi: 10.1136/ejhpharm-2025-004475. Online ahead of print.

ABSTRACT

Cefoperazone/sulbactam is a third-generation cephalosporin commonly used for severe infections. This case report presents a case of a 68-year-old woman who developed severe coagulation dysfunction and significant active bleeding after starting cefoperazone/sulbactam therapy following aortic dissection surgery. After discontinuing cefoperazone/sulbactam and administering vitamin K1, the patient's coagulation function returned to normal, with no further abnormalities after changing antibiotics. On assessing causality of the adverse drug reaction, the Naranjo scale for cefoperazone/sulbactam was 6. This case highlights the risks of cefoperazone/sulbactam in patients with underlying conditions such as renal insufficiency and malnutrition, emphasising the need for carefully monitoring coagulation parameters and dose adjustment to reduce the occurrence of drug-induced coagulopathy and healthcare-associated complications. Additionally, this case serves as a reminder of the vital contributions that clinical pharmacists make in monitoring and managing medication therapy, stressing the importance of fostering collaboration between clinical pharmacists and other healthcare workers.

PMID:40250970 | DOI:10.1136/ejhpharm-2025-004475

JCO Precis Oncol. 2025 Apr;9:e2400858. doi: 10.1200/PO-24-00858. Epub 2025 Apr 18.

ABSTRACT

PURPOSE: Contemporary prophylactic antiemetic regimens have improved the control of chemotherapy-induced nausea and vomiting (CINV). However, over 50% of patients still suffer from nausea. This study aimed to correlate the genetic determinants of individual patients with the efficacy of three prophylactic antiemetic regimens.

METHODS: Patients with breast cancer in two previously reported prospective antiemetic studies consented for the present pharmacogenetic study. Before high-emetogenic doxorubicin and cyclophosphamide (AC) (neo)adjuvant chemotherapy, they received a combination of antiemetic prophylaxis: regimen A and regimen B were, respectively, aprepitant/ondansetron/dexamethasone with or without olanzapine; regimen C was netupitant/palonosetron/dexamethasone. The effectiveness of antiemetic regimens was mainly assessed by complete protection (CP) rates. Patients' genotypes in three genes, HTR3A, HTR3B and TACR1, were analyzed.

RESULTS: Patients who were homozygous TT (p.129Tyr) of a non-nonsynonymous variant in HTR3B rs1176744 and homozygous GG of TACR1 rs3821313 had better outcome with regimen B. Digenic interaction analysis further reveals interaction between rs1176744 and rs3821313. Homozygotes TT of rs1176744 and homozygotes GG of rs3821313 achieved the highest CP rate with regimen B (10/12 patients; 83%), in contrast to only 29% (7/24) with regimen A (P = .0027). Homozygotes GG in both HTR3A rs1176722 and TACR1 rs3821313 showed the poorest response to regimen A with a CP rate of 17% (2/12), whereas patients given regimen B had the highest CP rate (70%; 7/10; P = .0159). The findings were confirmed upon logistic regression adjusted for clinical factors.

CONCLUSION: The present study confirmed our hypothesis that among Chinese patients with breast cancer who received AC, the selection of optimal antiemetic prophylaxis may be aided by assessing an individual's pharmacogenetic profile. It also highlights a novel digenic interaction that has not been known before for pharmacogenetic analysis.

PMID:40249884 | DOI:10.1200/PO-24-00858

Eur J Radiol. 2025 Apr 14;187:112115. doi: 10.1016/j.ejrad.2025.112115. Online ahead of print.

ABSTRACT

Intraluminal causes of small bowel obstruction (SBO) are less common than mural or extrinsic etiologies. This review categorizes intraluminal causes of SBO into four broad categories to provide a diagnostic framework for radiologic interpretation: 1) ingested contents, 2) bowel stasis, 3) inflammatory causes, and 4) neoplasms. Ingested materials can result in SBO when individual or accumulated contents are too large to pass, such as in the case of foreign bodies or bezoars. Bowel stasis causing SBO can be secondary to abnormal bowel function, such as in cystic fibrosis, reduced transit of contents at sites of narrowing such as surgical anastomoses, or the formation of enteroliths in diverticula which may subsequently dislodge and result in luminal obstruction. Inflammatory causes of SBO include strictures or fistulas that allow foreign bodies (such as gallstones) formed outside the bowel to enter the bowel lumen and cause obstruction. Finally, neoplasms can present as endophytic masses that occlude the bowel lumen through a ball-valve mechanism or serve as a lead point for intussusception. Recognizing the imaging features that are suggestive of intraluminal SBO is critical for accurate diagnosis and timely patient care.

PMID:40250005 | DOI:10.1016/j.ejrad.2025.112115

Comput Biol Med. 2025 Apr 17;191:110164. doi: 10.1016/j.compbiomed.2025.110164. Online ahead of print.

ABSTRACT

The human eye is a vital sensory organ that is crucial for visual perception. The retina is the main component of the eye and is responsible for visual signals. Due to its characteristics, the retina can reveal the occurrence of ocular diseases. So, early detection and automated diagnosis of retinal disease are crucial for preventing both temporary and permanent blindness. In the proposed work, a comprehensive framework is introduced, meticulously designed to leverage the synergic strengths of EfficientNet-B4 and Vision Transformers for attention-driven sophisticated analysis, offering a promising tool for advanced ophthalmic healthcare. This framework transcends the conventional hybridization by embedding the EfficientNetB4 reimagined as the multiscale feature encoder, creating discriminative feature maps preserving both local and intermediate contextual information. Then, Vision Transformer are incorporated to capitalize on the attention mechanisms to capture and model the global dependencies effectively. This combination establishes a sophisticated paradigm for capturing intricate patterns, focusing on the pertinent factors of the image, enabling precise and reliable classification. It is seen that the proposed model achieved a significant advancement by scoring an AUC of 0.9466, mAP of 0.7865, F1-score of 0.75 and Model Score of 0.8665. The framework achieved a remarkable 5.17% increase in the overall score when compared to the previous cutting-edge technologies on the same task. This improvement underscores the effectiveness of the hybrid model in identifying both local and global contextual information, making it a robust and reliable tool for precise retinal diagnosis.

PMID:40249994 | DOI:10.1016/j.compbiomed.2025.110164

Comput Biol Med. 2025 Apr 17;191:110166. doi: 10.1016/j.compbiomed.2025.110166. Online ahead of print.

ABSTRACT

Early detection of brain tumors in MRI images is vital for improving treatment results. However, deep learning models face challenges like limited dataset diversity, class imbalance, and insufficient interpretability. Most studies rely on small, single-source datasets and do not combine different feature extraction techniques for better classification. To address these challenges, we propose a robust and explainable stacking ensemble model for multiclass brain tumor classification. To address these challenges, we propose a stacking ensemble model that combines EfficientNetB0, MobileNetV2, GoogleNet, and Multi-level CapsuleNet, using CatBoost as the meta-learner for improved feature aggregation and classification accuracy. This ensemble approach captures complex tumor characteristics while enhancing robustness and interpretability. The proposed model integrates EfficientNetB0, MobileNetV2, GoogleNet, and a Multi-level CapsuleNet within a stacking framework, utilizing CatBoost as the meta-learner to improve feature aggregation and classification accuracy. We created two large MRI datasets by merging data from four sources: BraTS, Msoud, Br35H, and SARTAJ. To tackle class imbalance, we applied Borderline-SMOTE and data augmentation. We also utilized feature extraction methods, along with PCA and Gray Wolf Optimization (GWO). Our model was validated through confidence interval analysis and statistical tests, demonstrating superior performance. Error analysis revealed misclassification trends, and we assessed computational efficiency regarding inference speed and resource usage. The proposed ensemble achieved 97.81% F1 score and 98.75% PR AUC on M1, and 98.32% F1 score with 99.34% PR AUC on M2. Moreover, the model consistently surpassed state-of-the-art CNNs, Vision Transformers, and other ensemble methods in classifying brain tumors across individual four datasets. Finally, we developed a web-based diagnostic tool that enables clinicians to interact with the proposed model and visualize decision-critical regions in MRI scans using Explainable Artificial Intelligence (XAI). This study connects high-performing AI models with real clinical applications, providing a reliable, scalable, and efficient diagnostic solution for brain tumor classification.

PMID:40249992 | DOI:10.1016/j.compbiomed.2025.110166

J Med Internet Res. 2025 Apr 18;27:e66530. doi: 10.2196/66530.

ABSTRACT

BACKGROUND: Endometrial cancer is one of the most common gynecological tumors, and early screening and diagnosis are crucial for its treatment. Research on the application of artificial intelligence (AI) in the diagnosis of endometrial cancer is increasing, but there is currently no comprehensive meta-analysis to evaluate the diagnostic accuracy of AI in screening for endometrial cancer.

OBJECTIVE: This paper presents a systematic review of AI-based endometrial cancer screening, which is needed to clarify its diagnostic accuracy and provide evidence for the application of AI technology in screening for endometrial cancer.

METHODS: A search was conducted across PubMed, Embase, Cochrane Library, Web of Science, and Scopus databases to include studies published in English, which evaluated the performance of AI in endometrial cancer screening. A total of 2 independent reviewers screened the titles and abstracts, and the quality of the selected studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. The certainty of the diagnostic test evidence was evaluated using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) system.

RESULTS: A total of 13 studies were included, and the hierarchical summary receiver operating characteristic model used for the meta-analysis showed that the overall sensitivity of AI-based endometrial cancer screening was 86% (95% CI 79%-90%) and specificity was 92% (95% CI 87%-95%). Subgroup analysis revealed similar results across AI type, study region, publication year, and study type, but the overall quality of evidence was low.

CONCLUSIONS: AI-based endometrial cancer screening can effectively detect patients with endometrial cancer, but large-scale population studies are needed in the future to further clarify the diagnostic accuracy of AI in screening for endometrial cancer.

TRIAL REGISTRATION: PROSPERO CRD42024519835; https://www.crd.york.ac.uk/PROSPERO/view/CRD42024519835.

PMID:40249940 | DOI:10.2196/66530

JCO Clin Cancer Inform. 2025 Apr;9:e2400227. doi: 10.1200/CCI-24-00227. Epub 2025 Apr 18.

ABSTRACT

PURPOSE: Recurrences after curative resection in early-stage and locoregionally advanced non-small cell lung cancer (NSCLC) are common, necessitating a nuanced understanding of associated risk factors. This study aimed to establish a natural language processing (NLP) system to efficiently curate recurrence data in NSCLC and analyze risk factors longitudinally.

PATIENTS AND METHODS: Electronic health records of 6,351 patients with NSCLC with >700,000 notes were obtained from Mount Sinai's data sets. A deep learning-based customized NLP system was developed to identify cohorts experiencing recurrence. Recurrence types and rates over time were stratified by various clinical features. Cohort description analysis, Kaplan-Meier analysis for overall recurrence-free survival (RFS) and distant metastasis-free survival (DMFS), and Cox proportional hazards analysis were performed.

RESULTS: Of 1,295 patients with stage I-IIIA NSCLC with surgical resections, 336 patients (25.9%) experienced recurrence, as identified through NLP. The NLP system achieved a precision of 94.3%, a recall of 93%, and an F1 score of 93.5. Among 336 patients, 52.4% had local/regional recurrences, 44% distant metastases, and 3.6% unknown recurrence. RFS rates at years 1-5 were 93%, 81%, 73%, 67%, and 61%, respectively (96%, 89%, 84%, 80%, and 75% for distant metastasis). Stage-specific RFS rates at year 5 were 73% (IA), 62% (IB), 47% (IIA), 46% (IIB), and 20% (IIIA). Stage IB patients had a significantly higher likelihood of recurrence versus stage IA (adjusted hazard ratio [aHR], 1.63; P = .02). The RFS was lower in patients with clinically significant TP53 alteration (v TP53-negative or unknown significance), affecting overall RFS (aHR, 1.89; P = .007) and DMFS (aHR, 2.47; P = .009) among stage IA/IB patients.

CONCLUSION: Our scalable NLP system enabled us to generate real-world insights into NSCLC recurrences, paving the way for predictive models for preventing, diagnosing, and treating NSCLC recurrence.

PMID:40249880 | DOI:10.1200/CCI-24-00227