Chemistry. 2025 Aug 8:e02067. doi: 10.1002/chem.202502067. Online ahead of print.

ABSTRACT

The inherently low sensitivity of solution-state Nuclear Magnetic Resonance (NMR) has long limited its ability to characterize transient biomolecular states at atomic resolution. While dissolution dynamic nuclear polarization (dDNP) offers a compensating signal enhancement, its broader use has been hindered by rapid polarization decay, causing severe spectral distortion. Here, we introduce HyperW-Decon, an approach that enables high-sensitivity, high-resolution NMR of biomolecules in solution. HyperW-Decon combines two key aspects: (i) the use of hyperpolarized water (HyperW) to transfer polarization to proteins through rapid proton exchange; and (ii) a theory-driven, machine learning (ML)-based deconvolution method that corrects polarization-induced artifacts without requiring external reference signals. This approach is based on a first-principles understanding of dDNP line shapes and delivers a scalable solution to spectral distortion. Applied to intrinsically disordered proteins (IDPs) involved in biomineralization, HyperW-Decon reveals previously inaccessible, short-lived ion-peptide encounter complexes with residue resolution.

PMID:40778633 | DOI:10.1002/chem.202502067

Liver Int. 2025 Sep;45(9):e70255. doi: 10.1111/liv.70255.

ABSTRACT

BACKGROUND & AIMS: Immune checkpoint inhibitors (ICIs) are therapy for many malignancies including hepatocellular carcinoma (HCC), yet the impact of HCC on immune-mediated liver injury from checkpoint inhibitors (ILICI) remains poorly understood and no direct comparison exists for hepatotoxicity rates between ICI and sorafenib in HCC.

METHODS: In this retrospective cohort study, we extracted data on adult patients treated with five ICI regimens for HCC or non-HCC cancers, and HCC patients who received sorafenib between 2010 and 2020. The primary outcome was grade ≥ 3 ILICI or sorafenib (DILI). Logistic regression estimated adjusted odds ratios (OR) for liver injury.

RESULTS: We identified 530 patients, 129 (24%) HCC-ICI, 256 (48%) non-HCC ICI, and 145 (27%) HCC-sorafenib. Compared to non-HCC ICI, HCC-ICI and HCC-sorafenib were more often male (57%, 82%, 77%), Hispanic (14%, 35%, 34%), and cirrhotic (1%, 85%, 88%). Twenty-three patients developed grade ≥ 3 ILICI. ILICI incidence was higher for HCC-ICI (11%, CI 6-18) versus non-HCC ICI (4%, CI 2-6, p = 0.006) and DILI in HCC-sorafenib (3%, CI 1-8, p = 0.02) with incidence highest for ipilimumab-nivolumab (HCC-ICI 42%, CI 15-72 versus non-HCC 10%, CI 3-24; p = 0.02). On multivariable regression, ILICI was associated with HCC (OR 4.5, CI 1.8-11.4, p = 0.002) and treatment with ipilimumab-nivolumab (OR 6.9, CI 2.6-18.3, p < 0.001). Incidence of liver injury in HCC remained elevated for ICI versus sorafenib (OR 3.5, CI 1.2-10.4, p = 0.02).

CONCLUSIONS: We identified an elevated risk of liver injury in HCC patients receiving ICIs compared to ICI-treated non-HCC cancers and sorafenib-treated HCC, with dual ipilimumab-nivolumab therapy carrying the highest risk.

PMID:40778804 | DOI:10.1111/liv.70255

Pharmacol Res Perspect. 2025 Aug;13(4):e70164. doi: 10.1002/prp2.70164.

ABSTRACT

This study assessed whether the completeness of spontaneously reported adverse drug reaction (ADR) reports differs between consumers and healthcare professionals when submitted directly to regulators, and how this compares to reports from pharmaceutical companies. ADR reports (2014-2023) were obtained from public databases in Canada and the United Kingdom (UK), focusing on the medicine classes sodium-glucose cotransporter 2 inhibitors, glucagon-like peptide 1 receptor agonists, and dipeptidyl peptidase-4 inhibitors. ADR report completeness was assessed using vigiGrade tool variables. Descriptive statistics and chi-square tests were used for analysis. A total of 17 897 reports were analyzed-13 613 from the UK Yellow Card Scheme and 4284 from Canada. Most Canadian reports were submitted by pharmaceutical companies (55%), while in the UK, healthcare professionals submitted the majority (69%). Few reports were submitted directly by consumers in either Canada (4%) or the UK (7%). In Canada, the average completeness was 82% for consumer and healthcare professional reports and 57% for pharmaceutical companies. In the UK, completeness was 80% (consumers), 82% (healthcare professionals), and 69% (pharmaceutical companies). Canadian pharmaceutical company reports were significantly less complete for age, sex, outcome, dose, indication, and route of administration (all p < 0.001). In the UK, they were less complete for age, sex, and route of administration (all p < 0.001). In conclusion, reports submitted directly to regulators by consumers and healthcare professionals were more complete than those from pharmaceutical companies. The low consumer reporting rate, yet high completeness rate, highlights the need to encourage direct reporting to regulators to improve medicine safety monitoring.

PMID:40778745 | DOI:10.1002/prp2.70164

Stud Health Technol Inform. 2025 Aug 7;329:1942-1943. doi: 10.3233/SHTI251290.

ABSTRACT

This study develops a knowledge graph-based intelligent Q&A system for rare diseases, integrating data from 126 diseases, 2,609 symptoms, and related departments. The Bert-BiLSTM-CRF model achieved 82.13% accuracy in entity extraction, and TextCNN achieved 94.54% accuracy in intent recognition. The system improves access to medical knowledge but requires further optimization to reduce response times and handle a broader range of user queries.

PMID:40776307 | DOI:10.3233/SHTI251290

Stud Health Technol Inform. 2025 Aug 7;329:1054-1058. doi: 10.3233/SHTI251000.

ABSTRACT

Generating differential diagnoses for rare disease patients can be time intensive and highly dependent on the background and training of the evaluating physicians. Large language models (LLMs) have the potential to complement this process by automatically generating differentials to support physicians, but their performance in real-world patient populations remains underexplored. To this end, we assessed the diagnostic accuracy of ChatGPT-4o, Llama 3.1-8B-Instruct, and Exomiser in 424 rare disease patients at the Undiagnosed Diseases Network. ChatGPT-4o had the highest differential diagnostic accuracy (22.4% [95% CI: 18.4, 26.4]), outperforming Exomiser (13.9% [10.6, 17.2]; p < 0.001) and Llama 3.1-8B-Instruct (11.6% [8.5, 14.6]; p < 0.001). Adjusting for other factors, age at symptom onset was a significant predictor of ChatGPT-4o's diagnostic accuracy with the model performing better in patients with later symptom onset, potentially due to more distinct phenotypic presentations in older individuals. The combined accuracy of ChatGPT-4o and Exomiser was 30% [25.6, 34.3] and higher than that of either model alone (p < 0.01). This improvement highlights the potential of combining LLMs and bioinformatic models to generate differential diagnoses for rare diseases.

PMID:40776018 | DOI:10.3233/SHTI251000

Stud Health Technol Inform. 2025 Aug 7;329:683-687. doi: 10.3233/SHTI250927.

ABSTRACT

Developmental and Epileptic Encephalopathies (DEEs) are rare, severe conditions often discussed by families on social media, offering valuable insights into their experiences. Identifying these messages amidst unrelated content is crucial but challenging due to data imbalance. This study evaluates different uses of generative large language models (LLMs) for binary classification of DEE-related experiences within social media posts. Using CamemBERT as a baseline, we compared two strategies: zero-shot prompt-based classification and synthetic data generation for minority class augmentation. While zero-shot prompting underperformed, the addition of 2% synthetic data improved all metrics (macro/positive F1, precision and recall). Higher proportions of synthetic data led to decreased precision. These findings underscore the potential of hybrid approaches combining fine-tuning and domain-specific synthetic data for addressing data imbalance in rare disease contexts. Further validation across models and datasets is needed.

PMID:40775945 | DOI:10.3233/SHTI250927

Orphanet J Rare Dis. 2025 Aug 7;20(1):407. doi: 10.1186/s13023-025-03894-y.

ABSTRACT

BACKGROUND: Obtaining health-state utility values (HSUVs) aids in making scientific decisions in patient health management, especially for rare disease patients. However, there is currently no research specifically measuring the HSUVs of Chinese hemophilia B patients. Therefore, this study aims to assess the HSUVs of hemophilia B patients in China and explore its potential influencing factors.

METHODS: The sociodemographic characteristics of patients were obtained from the Beijing Hemophilia Home Care Center (BHHCC) database. And the HSUVs were further obtained by reaching hemophilia B patients through an application of BHHCC and the Chinese version of the EQ-5D-5L. The beta regression model was used to explore the potential influencing factors of the HSUVs of patients.

RESULTS: A total of 167 male patients (hemophilia B is an X-chromosome recessive disorder and female patients are rare) were included in the study. The mean age, HSUV and EQ-VAS were 20.01 ± 15.83, 0.755 ± 0.291 and 71.7 ± 22.7, respectively. The ceiling effects was 29.24%, and patients were more likely to experience problems in Pain/discomfort (57.49%). Compared to self-completion, proxy may overestimate HSUVs of patients. Pain (p < 0.000), disability (p < 0.000), complications (p < 0.001), inhibitors (p < 0.01), drug usage (p < 0.001), and bleeds (p < 0.000) were significantly associated with HSUVs in Chinese hemophilia B patients.

CONCLUSIONS: This study first assessed the HSUVs of Chinese hemophilia B patients, which provides support for further economic studies. Potential factors that affect the HSUVs of Chinese hemophilia B patients were also explored, which can provide a reference for developing health management measures. However, to enact more comprehensive and reliable disease management decisions, the effects of self-completion and proxy on the HSUVs of hemophilia B patients in China need to be further explored as well as the effects of specific factors.

PMID:40775730 | DOI:10.1186/s13023-025-03894-y

Orphanet J Rare Dis. 2025 Aug 7;20(1):410. doi: 10.1186/s13023-025-03933-8.

ABSTRACT

BACKGROUND: Rare diseases, though individually uncommon, collectively affect a significant portion of the population. However, their epidemiology in China remains underexplored. A population-based rare disease registry comprising 14.31 million individuals was conducted between 2012 and 2023 by the Beijing Municipal Health Big Data and Policy Research Center. Rare disease cases were identified via ICD-10 codes mapped to China's national rare disease lists (2018 and 2023) and international databases. Age-standardized incidence rates (ASIR) were calculated per 100,000 person-years with 95% confidence intervals.

RESULTS: Our analysis identified 12,371 rare disease cases, with the overall ASIR increasing from 6.109 in 2012 to 7.394 in 2023. Rare neurologic diseases accounted for 52.12% of cases, followed by systemic and rheumatologic diseases (16.89%) and rare neoplastic diseases (9.99%). The most frequently diagnosed rare diseases included generalized myasthenia gravis, ANCA-associated vasculitis, and malignant melanoma. Significant sex-based differences were observed, with female patients more affected by systemic and rheumatologic conditions, while male patients showed a higher incidence of respiratory disorders. Pediatric patients predominantly presented with inborn errors of metabolism and rare immune diseases. Comparisons with global data revealed notable disparities, such as a higher prevalence of Wilson's disease and a lower incidence of amyotrophic lateral sclerosis (ALS) in China.

CONCLUSIONS: This study represents the first large-scale, population-based analysis of rare diseases in China, revealing distinct epidemiological patterns. These findings underscore the critical need for healthcare policies that address the unique challenges posed by rare diseases in China.

PMID:40775651 | DOI:10.1186/s13023-025-03933-8

Stud Health Technol Inform. 2025 Aug 7;329:1594-1595. doi: 10.3233/SHTI251118.

ABSTRACT

Pharmacogenomics (PGx) has the potential to improve prescribing practices. In 2012, our institution developed a multidisciplinary program to implement PGx-CDS to support the clinical practice. CDS has been implemented for 45 drug-gene interactions, in 42,000 patients with PGx tests results, and in 2023, approximately 9281 CDS interventions have triggered at the time of prescribing. The program has been successful.

PMID:40776135 | DOI:10.3233/SHTI251118

Stud Health Technol Inform. 2025 Aug 7;329:1079-1083. doi: 10.3233/SHTI251005.

ABSTRACT

Pharmacogenomics (PGx) is pivotal in personalized medicine, particularly in cancer care, where drug efficacy and toxicity often vary with genetic variability. This study investigates the prevalence of CYP2D6-related medications and phenotypes in a cohort of 5,576 female breast cancer patients using genomic and electronic health record (EHR) data from the NIH's All of Us Research Program. A total of 77% of patients were prescribed at least one CYP2D6 metabolized drug. A customized pipeline was developed to determine CYP2D6 genotypes and phenotypes, identifying actionable phenotypes in 12.5% of patients. The prevalence of CYP2D6-associated drugs, including Tamoxifen, ondansetron, and tramadol, were widely prescribed and increased significantly following cancer diagnosis. Approximately 25% of phenotyped patients exhibited non-normal metabolizer types, emphasizing the importance of pharmacogenomic considerations in clinical decision-making. These findings emphasize the potential of incorporating PGx guidelines into routine clinical practice to optimize breast cancer treatment and improve patient outcomes.

PMID:40776023 | DOI:10.3233/SHTI251005

bioRxiv [Preprint]. 2025 Jul 21:2025.07.21.666023. doi: 10.1101/2025.07.21.666023.

ABSTRACT

BACKGROUND: Cystic fibrosis (CF) is caused by mutations in the CFTR gene, leading to epithelial dysfunction and progressive lung disease. Although CFTR modulators have transformed care, ∼10% of people with CF remain without effective therapy. Durable, mutation-agnostic approaches are urgently needed.

METHOD: We used a lentiviral (LV) vector to deliver wild-type CFTR to airway basal cells derived from 13 paediatric CF participants with a range of genotypes. Transduced cells were assessed for transgene expression, epithelial differentiation, and CFTR function using air-liquid interface (ALI) cultures. Separately, to evaluate regenerative capacity in vivo , LV GFP -transduced rabbit airway basal cells were transplanted into the denuded nasal septum of healthy New Zealand white rabbits using a biocompatible scaffold.

RESULTS: Transduced basal cells retained multilineage differentiation capacity, forming well-organized, pseudostratified epithelium with intact barrier function and ciliary activity. CFTR channel activity was restored to levels comparable to or exceeding those achieved with elexacaftor/tezacaftor/ivacaftor (ETI), including in individuals with nonsense mutations. Combined CFTR transduction plus ETI treatment showed additive benefit. In vivo , transplanted rabbit basal cells engrafted and differentiated to regenerate a mucociliary epithelium, with improved nasal potential difference and mucociliary clearance compared to scaffold-only controls.

CONCLUSION: Our study demonstrates that LV-mediated CFTR gene addition restores CFTR function in vitro across genotypes and supports epithelial regeneration in a clinically relevant animal airway model. This two-part platform offers a scalable path toward cell therapies for all people with CF and may have broader applications in upper airway epithelial repair.

PMID:40777307 | PMC:PMC12330628 | DOI:10.1101/2025.07.21.666023

Data Brief. 2025 Jul 16;62:111890. doi: 10.1016/j.dib.2025.111890. eCollection 2025 Oct.

ABSTRACT

This dataset presents a comprehensive multimodal collection of data acquired from the chewing of beef and plant-based burger patties using both human participants (Invivo) and a biomimicking robotic chewing device (Invitro). The primary objective of the data collection was to discover relationships regarding the change in food bolus properties with the number of robotic chewing cycles as the human swallowing threshold is achieved, which will facilitate the development of deep learning models capable of predicting mechanical and textural properties of chewed food boluses from images. In the in-vivo experiments, expectorated bolus samples were collected from three healthy adult male participants, who chewed food samples until just before swallowing. The chewed boluses were then imaged using a 12MP camera and a flatbed scanner, followed by Texture Profile Analysis (TPA) to measure texture parameters. The dataset comprises two main folders Invivo and Invitro. The Invivo data thus comprises high-resolution images and corresponding TPA metrics at the near-swallowing stage. In the Invitro experiments, a 3 degree of freedom linkage chewing robot (ChewBot) with a soft robotic oral cavity was used to simulate human mastication. The robot performed controlled mastication using different molar trajectories that varied in lateral shearing effect. Food samples were chewed for up to 40 chewing cycles, with artificial saliva introduced at 10 % of the food's weight. For each experimental condition, the dataset includes real-time images captured immediately after each the robotic chewing cycle, force profiles recorded at 100 ms intervals during the chewing, and TPA metrics of the resulting bolus after every 5 chewing cycles. This dataset has significant reuse potential in various fields. In food science, it can support studies on the mechanical breakdown of meat and meat alternatives, aiding in the reformulation of plant-based foods to better mimic desirable animal-based food textures. This dataset supports rehabilitation in health sciences by aiding personalized diet design for individuals with jaw disorders or dysphagia and guiding texture-appropriate menu options for patients. In robotics and artificial mastication, it informs the development of chewing systems. It also enables machine learning applications for predicting food texture from images, allowing automated, non-invasive analysis.

PMID:40778379 | PMC:PMC12329220 | DOI:10.1016/j.dib.2025.111890

Ophthalmol Sci. 2025 Jun 19;5(6):100861. doi: 10.1016/j.xops.2025.100861. eCollection 2025 Nov-Dec.

ABSTRACT

CLINICAL RELEVANCE: Corneal ulcers cause preventable blindness in >2 million individuals annually, primarily affecting low- and middle-income countries. Prompt and accurate pathogen identification is essential for targeted antimicrobial treatment, yet current diagnostic methods are costly and slow and require specialized expertise, limiting accessibility.

METHODS: We systematically reviewed literature published from 2017 to 2024, identifying 37 studies that developed or validated artificial intelligence (AI) models for pathogen detection and related classification tasks in infectious keratitis. The studies were analyzed for model types, input modalities, datasets, ground truth determination methods, and validation practices.

RESULTS: Artificial intelligence models demonstrated promising accuracy in pathogen detection using image interpretation techniques. Common limitations included limited generalizability, lack of diverse datasets, absence of multilabeled classification methods, and variability in ground truth standards. Most studies relied on single-center retrospective datasets, limiting applicability in routine clinical practice.

CONCLUSIONS: Artificial intelligence shows significant potential to improve pathogen detection in infectious keratitis, enhancing both diagnostic accuracy and accessibility globally. Future research should address identified limitations by increasing dataset diversity, adopting multilabel classification, implementing prospective and multicenter validations, and standardizing ground truth definitions.

FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

PMID:40778364 | PMC:PMC12329105 | DOI:10.1016/j.xops.2025.100861

Ophthalmol Sci. 2025 Jun 16;5(6):100852. doi: 10.1016/j.xops.2025.100852. eCollection 2025 Nov-Dec.

ABSTRACT

PURPOSE: We present an algorithm to segment subretinal fluid (SRF) on individual B-scan slices in patients with rhegmatogenous retinal detachment (RRD). Particular attention is paid to robustness, with a fivefold cross-validation approach and a hold-out test set.

DESIGN: Retrospective, cross-sectional study.

PARTICIPANTS: A total of 3819 B-scan slices across 98 time points from 45 patients were used in this study.

METHODS: Subretinal fluid was segmented on all scans. A base SegFormer model, pretrained on 4 massive data sets, was further trained on raw B-scans from the retinal OCT fluid challenge data set of 4532 slices: an open data set of intraretinal fluid, SRF, and pigment epithelium detachment. When adequate performance was reached, transfer learning was used to train the model on our in-house data set, to segment SRF by generating a pixel-wise mask of presence/absence of SRF. A fivefold cross-validation approach was used, with an additional hold-out test set. All folds were first trained and cross-validated and then additionally tested on the hold-out set. Mean (averaged across images) and total (summed across all pixels, irrespective of image) Dice coefficients were calculated for each fold.

MAIN OUTCOME MEASURES: Subretinal fluid volume after surgical intervention for RRD.

RESULTS: The average total Dice coefficient across the validation folds was 0.92, the average mean Dice coefficient was 0.82, and the median Dice was 0.92. For the test set, the average total Dice coefficient was 0.94, the average mean Dice coefficient was 0.82, and the median Dice was 0.92. The model showed strong interfold consistency on the hold-out set, with a standard deviation of only 0.03.

CONCLUSIONS: The SegFormer model for SRF segmentation demonstrates a strong ability to segment SRF. This result holds up to cross-validation and hold-out testing, across all folds. The model is available open-source online.

FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

PMID:40778358 | PMC:PMC12329092 | DOI:10.1016/j.xops.2025.100852

Front Endocrinol (Lausanne). 2025 Jul 24;16:1634875. doi: 10.3389/fendo.2025.1634875. eCollection 2025.

ABSTRACT

OBJECTIVE: This study aimed to evaluate the value of constructing a multimodal deep-learning video model based on 2D ultrasound and contrast-enhanced ultrasound (CEUS) dynamic video for the preoperative prediction of OLNM in papillary thyroid carcinoma (PTC) patients.

METHODS: A retrospective analysis was conducted on 396 cases of clinically lymph node-negative PTC cases with ultrasound images collected between January and September 2023. Five representative deep learning architectures were pre-trained to construct deep learning static image models (DL_image), CEUS dynamic video models (DL_CEUSvideo), and combined models (DL_combined). The area under the receiver operating characteristic curve (AUC) was used to evaluate model performance, with comparisons made using the Delong test. A P-value of less than 0.05 was considered statistically significant.

RESULTS: The DL_CEUSvideo, DL_image, and DL_combined models were successfully developed and demonstrated. The AUC values were 0.826 (95% CI: 0.771-0.881), 0.759 (95% CI: 0.690-0.828), and 0.926 (95% CI: 0.891-0.962) in the training set, and 0.701 (95% CI: 0.589-0.813), 0.624 (95% CI: 0.502-0.745), and 0.734 (95% CI: 0.627-0.842) in the test set. Finally, sensitivity, specificity, and accuracy for the DL_CEUSvideo, DL_image, and DL_combined models were 0.836, 0.671, 0.704; 0.673, 0.716, 0.707; and 0.818, 0.902, 0.886 in the training set, and 0.556, 0.775, 0.724; 0.556, 0.674, 0.647; and 0.704, 0.663, 0.672 in the test set, respectively.

CONCLUSION: These results demonstrated that the multimodal deep learning dynamic video model could preoperatively predict OLNM in PTC patients. The DL_CEUSvideo model outperformed the DL_image model, while the DL_combined model significantly enhanced sensitivity without compromising specificity.

PMID:40778281 | PMC:PMC12329689 | DOI:10.3389/fendo.2025.1634875

Front Cell Dev Biol. 2025 Jul 24;13:1608580. doi: 10.3389/fcell.2025.1608580. eCollection 2025.

ABSTRACT

OBJECTIVE: To enhance the automatic detection precision of diabetic retinopathy (DR) lesions, this study introduces an improved YOLOv8 model specifically designed for the precise identification of DR lesions.

METHOD: This study integrated two attention mechanisms, convolutional exponential moving average (convEMA) and convolutional simple attention module (convSimAM), into the backbone of the YOLOv8 model. A dataset consisting of 3,388 ultra-widefield (UWF) fundus images obtained from patients with DR, each with a resolution of 2,600 × 2048 pixels, was utilized for both training and testing purposes. The performances of the three models-YOLOv8, YOLOv8+ convEMA, and YOLOv8+ convSimAM-were systematically compared.

RESULTS: A comparative analysis of the three models revealed that the original YOLOv8 model suffers from missed detection issues, achieving a precision of 0.815 for hemorrhage spot detection. YOLOv8+ convEMA improved hemorrhage detection precision to 0.906, while YOLOv8+ convSimAM achieved the highest value of 0.910, demonstrating the enhanced sensitivity of spatial attention. The proposed model also maintained comparable precision in detecting hard exudates while improving recall to 0.804. It demonstrated the best performance in detecting cotton wool spots and the epiretinal membrane. Overall, the proposed method provides a fine-tuned model specialized in subtle lesion detection, providing an improved solution for DR lesion assessment.

CONCLUSION: In this study, we proposed two attention-augmented YOLOv8 models-YOLOv8+ convEMA and YOLOv8+ convSimAM-for the automated detection of DR lesions in UWF fundus images. Both models outperformed the baseline YOLOv8 in terms of detection precision, average precision, and recall. Among them, YOLOv8+ convSimAM achieved the most balanced and accurate results across multiple lesion types, demonstrating an enhanced capability to detect small, low-contrast, and structurally complex features. These findings support the effectiveness of lightweight attention mechanisms in optimizing deep learning models for high-precision DR lesion detection.

PMID:40778265 | PMC:PMC12328430 | DOI:10.3389/fcell.2025.1608580

Korean J Neurotrauma. 2025 Jul 18;21(3):172-182. doi: 10.13004/kjnt.2025.21.e22. eCollection 2025 Jul.

ABSTRACT

Vertebral fractures are prevalent skeletal injuries commonly associated with osteoporosis, trauma, and degenerative diseases. Early and accurate diagnosis is crucial to prevent complications such as chronic pain and progressive spinal deformities. In recent years, artificial intelligence (AI) has emerged as a powerful tool in medical imaging to support automatic detection and classification of vertebral fractures. This review provides an overview of AI-based approaches for spinal fracture diagnosis and summarizes recent advances in deep learning (DL) and machine learning (ML) models. The performance of AI models, mainly evaluated by sensitivity, specificity, and accuracy metrics, varies with imaging modality and dataset size, with computed tomography-based models demonstrating superior diagnostic accuracy. In addition, AI-assisted workflows have been shown to improve diagnostic efficiency, reducing the time required for fracture detection. Despite these advances, challenges remain, such as dataset variability, the need for large-scale annotated datasets, and standardization of evaluation metrics. Future research should focus on improving model generalization, integrating multimodal imaging data, and validating AI applications in real-world clinical settings to further improve vertebral fracture diagnosis and patient management.

PMID:40778250 | PMC:PMC12325887 | DOI:10.13004/kjnt.2025.21.e22

medRxiv [Preprint]. 2025 Jul 21:2024.09.23.24314186. doi: 10.1101/2024.09.23.24314186.

ABSTRACT

Dementia is a complex condition whose multifaceted nature poses significant challenges in the diagnosis, prognosis, and treatment of patients. Despite the availability of large open-source data fueling a wealth of promising research, effective translation of preclinical findings to clinical practice remains difficult. This barrier is largely due to the complexity of unstructured and disparate preclinical and clinical data, which traditional analytical methods struggle to handle. Novel analytical techniques involving Deep Learning (DL), however, are gaining significant traction in this regard. Here, we have investigated the potential of a cascaded multimodal DL-based system (TelDem), assessing the ability to integrate and analyze a large, heterogeneous dataset (n=7,159 patients), applied to three clinically relevant use cases. Using a Cascaded Multi-Modal Mixing Transformer (CMT), we assessed TelDem's validity and (using a Cross-Modal Fusion Norm - CMFN) model explainability in (i) differential diagnosis between healthy individuals, AD, and three sub-types of frontotemporal lobar degeneration (ii) disease staging from healthy cognition to mild cognitive impairment (MCI) and AD, and (iii) predicting progression from MCI to AD. Our findings show that the CMT enhances diagnostic and prognostic accuracy when incorporating multimodal data compared to unimodal modeling and that cerebrospinal fluid (CSF) biomarkers play a key role in accurate model decision making. These results reinforce the power of DL technology in tapping deeper into already existing data, thereby accelerating preclinical dementia research by utilizing clinically relevant information to disentangle complex dementia pathophysiology.

PMID:40778154 | PMC:PMC12330412 | DOI:10.1101/2024.09.23.24314186

Smart Health (Amst). 2024 Sep;33:100498. doi: 10.1016/j.smhl.2024.100498. Epub 2024 Jun 5.

ABSTRACT

This work focuses on the development and manufacturing of a wireless, wearable, low-power fall detection sensor (FDS) designed to predict and detect falls in elderly at-risk individuals. Unintentional falls are a significant risk in this demographic, often resulting from diminished physical capabilities such as reduced hand grip strength and complications from conditions like arthritis, vertigo, and neuromuscular issues. To address this, we utilize advanced low-power field-programmable gate arrays (FPGAs) to implement a fixed-function neural network capable of categorizing activities of daily life (ADLs), including the detection of falls. This system employs a Convolutional Neural Network (CNN) model, trained and validated using the Caffe deep learning framework with data collected from human subjects experiments. This system integrates an ST Microelectronics LSM6DSOX inertial measurement unit (IMU) sensor, embedded with an ultra-low-power Lattice iCE40UP FPGA, which samples and stores joint acceleration and orientation rate. Additionally, we have acquired and published a dataset of 3D accelerometer and gyroscope measurements from predefined ADLs and falls, using volunteer human subjects. This innovative approach aims to enhance the safety and well-being of older adults by providing timely and accurate fall detection and prediction. In this paper, we present an innovative approach to utilizing a compact Convolutional Neural Network (CNN) core for accelerating convolutional operations on a machine learning model, suitable for deployment on an ultra-low power FPGA.

PMID:40777999 | PMC:PMC12327353 | DOI:10.1016/j.smhl.2024.100498

Digit Health. 2025 Aug 6;11:20552076251366855. doi: 10.1177/20552076251366855. eCollection 2025 Jan-Dec.

ABSTRACT

OBJECTIVES: Accurate segmentation of medical images is vital for effective disease diagnosis and treatment planning. This is especially important in resource-constrained environments. This study aimed to evaluate the performance of various U-Net-based deep learning architectures for chest X-ray (CXR) segmentation and identify the most effective model in terms of both accuracy and computational efficiency.

METHODS: We assessed the segmentation performance of eight U-Net variants: U-Net7, U-Net9, U-Net11, U-Net13, U-Net16, U-Net32, U-Net64, and U-Net128. The evaluation was conducted using a publicly available CXR dataset categorized into normal, COVID-19, and viral pneumonia classes. Each image was paired with a corresponding segmentation mask. Image preprocessing involved resizing, noise filtering, and normalization to standardize input quality. All models were trained under identical experimental conditions to ensure a fair comparison. Performance was evaluated using two key metrics: Intersection over Union (IoU) and Dice Coefficient (DC). Additionally, computational efficiency was measured by comparing the total number of trainable parameters and the training time for each model.

RESULTS: U-Net9 achieved the highest performance among all tested models. It recorded a DC of 0.98 and an IoU of 0.96, outperforming both shallower and deeper U-Net architectures. Models with increased depth or filter width, such as U-Net128, showed diminishing returns in accuracy. These models also incurred significantly higher computational costs. In contrast, U-Net16 and U-Net32 demonstrated reduced segmentation accuracy compared to U-Net9. Overall, U-Net9 provided the optimal balance between precision and computational efficiency for CXR segmentation tasks.

CONCLUSION: The U-Net9 architecture offers a superior solution for CXR image segmentation. It combines high segmentation accuracy with computational practicality, making it suitable for real-world applications. Its implementation can support radiologists by enabling faster and more reliable diagnoses. This can lead to improved clinical decision-making and reduced diagnostic delays. Future work will focus on integrating U-Net9 with multimodal imaging data, such as combining CXR with computerized tomography or MRI scans. Additionally, exploration of advanced architectures, including attention mechanisms and hybrid models, is planned to further enhance segmentation performance.

PMID:40777837 | PMC:PMC12329272 | DOI:10.1177/20552076251366855