Papillary Thyroid Carcinoma (PTC) accounts for more than 80% of thyroid cancers, and cervical Central Lymph Node Metastasis (CLNM) is a critical risk factor affecting patient prognosis. Conventional ultrasound is the first-line method for preoperative evaluation of PTC, but it is restricted by operator subjectivity, image quality interference, and limited diagnostic accuracy for CLNM. Multimodal ultrasound radiomics and deep learning have emerged as innovative technical approaches to address these limitations. This paper reviews the research progress of multimodal ultrasound radiomics and deep learning in predicting CLNM in PTC, covering modality composition, feature extraction, multimodal fusion strategies, model efficacy, and clinical application value. Current studies confirm that multimodal ultrasound radiomics integrating B-mode, contrast-enhanced ultrasound, elastography and microflow imaging achieves an Area Under the Curve (AUC) of 0.78–0.97 for CLNM prediction, significantly outperforming single-modal ultrasound. Deep learning models based on convolutional neural networks and Transformer architectures realize end-to-end automatic prediction with high efficiency. However, most studies are retrospective single-center designs, resulting in insufficient model generalization; the "black box" problem of artificial intelligence and the shallow application of Explainable Artificial Intelligence (XAI) also restrict clinical translation. In conclusion, multimodal ultrasound radiomics and deep learning break through the bottlenecks of conventional ultrasound and provide reliable support for preoperative non-invasive risk assessment of PTC. Future research should focus on prospective multicenter trials, unified imaging standards, optimized XAI methods and multi-omics integration to promote clinical application.

Tenosynovitis is a prevalent and debilitating musculoskeletal disorder with a rising incidence linked to metabolic diseases, an aging population, and modern lifestyle factors such as prolonged digital device use. Current diagnostic methods are limited by operator dependency and high costs, while pharmacological treatments are associated with adverse effects and limited long-term efficacy. To address these gaps, this study integrated Genome-Wide Association Studies (GWAS), published literature, and GEO RNA-seq data (GSE93698) to identify novel biomarkers and therapeutic targets. Differential expression analysis revealed that while most GWAS-derived genes showed no significant changes, IL6 and IGFBP3 were significantly upregulated, and CREB5 was downregulated in tenosynovitis. IGFBP3 emerged as a promising candidate given its role in regulating IGF-mediated cellular processes and its interaction with inflammatory pathways. Molecular dynamics simulations further demonstrated that Salvianolic Acid B (SAB) binds to the active pocket of IGFBP3, and structure-based chemical modification of the SAB catechol ring significantly enhanced binding affinity. These findings identify IGFBP3 as a potential early detection biomarker and a novel drug target for tenosynovitis, providing a foundation for the development of targeted, mechanism-based therapies.

Against the backdrop of the rapid development of medical imaging technology, a variety of mature imaging examination methods have been widely applied in clinical practice. Single Photon Emission Computed Tomography-Computed Tomography (SPECT-CT) features the integration of functional imaging and anatomical imaging, exhibiting unique advantages in the early diagnosis of diseases, precise localization of lesions, and therapeutic effect evaluation. However, in the process of clinical promotion of SPECT-CT, practical problems still exist, including immature quantitative image analysis, inconsistent radiation doses and diagnostic criteria, and high equipment costs, which objectively restrict its popularization in primary medical institutions and the full exertion of its diagnostic efficacy. On this basis, this paper systematically elaborates on the basic concepts, technical principles and development history of SPECT-CT imaging technology, with a focus on sorting out its current clinical application status in neurological diseases, cardiovascular diseases, tumor diseases, intrahepatic biliary system diseases and musculoskeletal system diseases. Furthermore, it conducts comprehensive analysis, summary and prospect outlook, aiming to provide evidence-based references for the standardized application and subsequent innovative development of this technology.

Objective: To thoroughly analyze the correlation of thrombus volume with admission neurological deficit severity and postoperative clinical outcomes in patients with Acute Ischemic Stroke (AIS) treated by mechanical thrombectomy. Methods: A total of 119 AIS patients receiving mechanical thrombectomy were retrospectively enrolled. According to the 90-day mRS score after surgery, subjects were divided into a favorable prognosis group (n = 74) and an unfavorable prognosis group (n = 45). Baseline clinical data of the two groups were comprehensively collected, including age, gender, history of hypertension, diabetes mellitus, coronary heart disease, underlying heart disease, admission NIHSS score, onset-to-treatment time, vascular occlusion site, postoperative intracranial complications and thrombus volume. Spearman correlation analysis was adopted to evaluate the correlation between thrombus volume and NIHSS score, and multivariate Logistic regression analysis was performed to identify whether thrombus volume acted as an independent predictor of favorable prognosis. Results: Patients in the unfavorable prognosis group had significantly higher age, admission NIHSS scores, longer onset-to-treatment time, larger thrombus volume and higher incidence of intracranial hemorrhage complications compared with the favorable prognosis group (all p < 0.05). There were no statistically significant intergroup differences in gender, history of hypertension, diabetes mellitus, coronary heart disease, underlying heart disease and distribution of vascular occlusion sites (all p > 0.05). Conclusion: Thrombus volume is strongly correlated with the severity of admission neurological deficits and long-term clinical outcomes in AIS patients treated with mechanical thrombectomy.

Human Serum Albumin (HSA) is the most abundant protein in human plasma and plays essential physiological roles, including the maintenance of colloid osmotic pressure and the transport of a wide range of endogenous and exogenous substances. Recombinant Human Serum Albumin (rHSA), produced through genetic engineering technologies, effectively overcomes the inherent limitations of plasma-derived Human Serum Albumin (pHSA) in terms of raw material availability and biosafety. Consequently, it has attracted sustained attention in the biopharmaceutical field. This review systematically summarizes the molecular basis, production strategies, and current applications of rHSA, with particular emphasis on comparing the technical characteristics and applicability of different expression systems. In addition, future directions for process innovation and industrial translation are discussed, with the aim of providing references for further research and development in this area.

Alzheimer's Disease (AD) is a multifactorial neurodegenerative disorder affecting over 55 million individuals worldwide, characterised by Amyloid beta (Aβ) plaques, neurofibrillary tangles, and sustained neuroinflammation. While amyloid- and tau-targeted therapies have dominated therapeutic research, their limited clinical efficacy has intensified focus on neuroinflammation as a central and modifiable disease mechanism. This review synthesises current understanding of neuroinflammatory pathogenesis in AD, with emphasis on Microglial polarisation (M1/M2), Disease-Associated Microglia (DAM), TREM2 signalling, and reactive astrocyte conversion. This paper further evaluates pharmacological strategies targeting these pathways, including cytokine inhibitors (TNF-α and IL-6 blockade), microglial modulators (CSF-1R inhibitors, TREM2 agonistic antibodies), and emerging innate immune targets (cGAS-STING pathway inhibitors, and S-palmitoylation inhibitors). Despite strong preclinical rationale, clinical translation has been impeded by Blood-Brain Barrier (BBB) penetration challenges, intervention timing, peripheral immunosuppression risks, and the biological redundancy of neuroimmune networks. Future therapeutic success will likely require combination approaches, CNS-targeted delivery systems, and biomarker-guided patient stratification to fully exploit the therapeutic potential of neuroinflammation-directed strategies in AD.

Clinical biostatistics has traditionally relied on regression-based models to explain associations, estimate risks, and support medical decision-making. The growth of electronic health records, imaging data, omics data, and follow-up data has made clinical information larger, less tidy, and more difficult to model with only classical methods. Machine learning is increasingly used in this setting because it can capture non-linear patterns and handle high-dimensional predictors. This narrative review summarizes applications of supervised learning, unsupervised learning, and deep learning in clinical diagnosis, prognosis prediction, patient stratification, and biomarker discovery. Literature was selected from PubMed, Web of Science, and Google Scholar, with emphasis on studies and reporting guidelines published from 2019 to 2025. This review finds that machine learning is useful when the clinical question is clear, data quality is acceptable, and validation is strict. However, a stronger algorithm does not automatically become a better clinical tool. Main limitations include weak interpretability, biased training data, poor transportability, overreliance on Area Under the Curve (AUC), and incomplete reporting. Machine learning should therefore be treated as a complement to traditional biostatistics rather than a simple replacement.