Multi-omics signatures
Several studies combine gene-expression datasets with computational modeling to derive diagnostic or prognostic signatures. In bladder cancer, an 18-gene glycolysis-associated signature was built with LASSO Cox regression and tested in TCGA, GEO, and IMvigor210 cohorts, while in Alzheimer’s disease, bulk RNA-seq and machine learning identified RBM3, GOLGA8A, ALS2, FSD2, and LOC100287628, which were further examined with single-cell RNA-seq and immune-infiltration analysis (PMIDs 39867879, 40190307). Other studies used pyroptosis-related genes in hepatocellular carcinoma and platelet-related genes TRIM22, TIMP1, and RAC2 in ulcerative colitis to connect molecular signatures with immune-cell infiltration, disease pathways, and clinical prediction (PMIDs 40331560, 42668808).
📄 Frontiers in immunology ↗📄 Archives of medical science : AMS ↗📄 Technology and health care : official journal of the European Society for Engineering and Medicine ↗📄 PeerJ ↗
Liquid biopsy profiling
Blood and urine samples, including extracellular vesicles, were investigated for disease biomarkers or treatment-response prediction. UHPLC-MS urine metabolomics generated separate panels for distinguishing major depressive disorder and bipolar disorder, while plasma exosomal microRNA profiling combined with machine learning was used to predict response to gemcitabine-based preoperative treatment in pancreatic cancer (PMIDs 38548210, 39278808). Other studies identified differentially expressed exosomal microRNAs in blood and urine from patients with renal interstitial fibrosis, developed an LNA-probe ddPCR assay for low-abundance plasma miR-192-5p in hepatocellular carcinoma, and built a one-step urinary extracellular-vesicle capture/SERS platform for urologic-cancer classification (PMIDs 39787458, 41167182, 41964667).
📄 Journal of affective disorders ↗📄 Pancreatology : official journal of the International Association of Pancreatology (IAP) ... [et al.] ↗📄 Indian journal of pathology & microbiology ↗📄 Clinical chemistry and laboratory medicine ↗📄 Analytical chemistry ↗
Clinical risk prediction
Several studies evaluated whether circulating, urinary, or immune measurements can predict outcomes or treatment response. In immune thrombocytopenia, antiplatelet autoantibodies, cytotoxic T-lymphocyte percentage, and the CD4+/CD8+ immunoregulatory index were assessed in relation to treatment responses during 12 months of follow-up; copeptin was measured before and after pituitary surgery to predict postoperative sodium disturbances (PMIDs 38251842, 38551356). Fecal calprotectin was evaluated for prediction of death or liver transplantation in advanced chronic liver disease, urinary renal tubular epithelial cells and casts were associated with renal outcomes in biopsy-proven diabetic nephropathy, and a plasma CER24 risk score was tested for mortality and nonfatal myocardial infarction in patients with known or suspected coronary artery disease (PMIDs 39031494, 39031241, 40685265).
📄 Hematology (Amsterdam, Netherlands) ↗📄 Neurosurgery ↗📄 United European gastroenterology journal ↗📄 Journal of nephrology ↗📄 Nutrition, metabolism, and cardiovascular diseases : NMCD ↗
Immune and inflammatory profiles
Serial plasma cytokine and chemokine measurements from patients with large B-cell lymphoma undergoing anti-CD19 CAR-T therapy were analyzed with machine-learning models for treatment response at 3 months, cytokine release syndrome, and immune effector cell-associated neurotoxicity syndrome, while a separate prospective pilot study combined wearable monitoring with peripheral-blood cytokine profiling to detect cytokine release syndrome in patients with relapsed or refractory multiple myeloma receiving CAR-T therapy (PMIDs 40113544, 42084945). In post-COVID-19 cohorts, endothelial-colony-forming cells and angiogenesis-related proteins were assessed at 6 and 12 months, while salivary cortisol was sampled at multiple times across the day to assess hypothalamic-pituitary-adrenal axis activity in long COVID (PMIDs 38324145, 41568000). Proteomic and transcriptomic profiling of subacute cutaneous lupus erythematosus identified IFI44 and EPSTI1, whose serum levels were measured across healthy, subacute cutaneous lupus erythematosus, systemic lupus erythematosus, and lupus nephritis groups; serum levels also differentiated subacute cutaneous lupus erythematosus from systemic lupus erythematosus (PMID 40323689).
📄 Clinical lymphoma, myeloma & leukemia ↗📄 JCI insight ↗📄 Infection ↗📄 Frontiers in cellular and infection microbiology ↗📄 Annals of medicine ↗
Imaging and digital biomarkers
Imaging and computational analysis were used to derive potential or prognostic biomarkers from spinal cord, pituitary, prostate cancer, and brain features. Diffusion tensor imaging and diffusion basis spectrum imaging metrics at the rostral C3 spinal level were related to two-year clinical outcomes after decompressive surgery for cervical spondylotic myelopathy, while MRI radiomic features and a LASSO-derived signature were used to distinguish posterior pituitary tumors from non-functional pituitary neuroendocrine tumors before surgery (PMIDs 38904404, 40229520). Digital histopathology combined with clinical variables produced a multimodal artificial-intelligence score that was associated with overall survival, clinical progression, and castration-resistant prostate cancer in metastatic hormone-sensitive prostate cancer (PMID 39665917). Resting-state fMRI features were selected to train machine-learning models distinguishing trigeminal neuralgia patients, including those with classical and idiopathic trigeminal neuralgia, from pain-free controls (PMID 40153851).
📄 Neurosurgery ↗📄 Journal of imaging informatics in medicine ↗📄 European urology oncology ↗📄 Journal of neurosurgery ↗