Machine-assisted pathological recognition has been focused on supervised learning (SL) that suffers from a significant annotation bottleneck. We propose a semi-supervised learning (SSL) method based on the mean teacher architecture using 13,111 whole slide images of colorectal cancer from 8803 subjects from 13 independent centers. SSL (~3150 labeled, ~40,950 unlabeled; ~6300 labeled, ~37,800 unlabeled patches) performs significantly better than the SL. No significant difference is found between SSL (~6300 labeled, ~37,800 unlabeled) and SL (~44,100 labeled) at patch-level diagnoses (area under the curve (AUC): 0.980 ± 0.014 vs. 0.987 ± 0.008, P value = 0.134) and patient-level diagnoses (AUC: 0.974 ± 0.013 vs. 0.980 ± 0.010, P value = 0.117), which is close to human pathologists (average AUC: 0.969). The evaluation on 15,000 lung and 294,912 lymph node images also confirm SSL can achieve similar performance as that of SL with massive annotations. SSL dramatically reduces the annotations, which has great potential to effectively build expert-level pathological artificial intelligence platforms in practice.

Although the gut microbiota has been reported to influence osteoporosis risk, the individual species involved, and underlying mechanisms, remain largely unknown. We performed integrative analyses in a Chinese cohort of peri-/post-menopausal women with metagenomics/targeted metabolomics/whole-genome sequencing to identify novel microbiome-related biomarkers for bone health. Bacteroides vulgatus was found to be negatively associated with bone mineral density (BMD), which was validated in US white people. Serum valeric acid (VA), a microbiota derived metabolite, was positively associated with BMD and causally downregulated by B. vulgatus. Ovariectomized mice fed B. vulgatus demonstrated increased bone resorption and poorer bone micro-structure, while those fed VA demonstrated reduced bone resorption and better bone micro-structure. VA suppressed RELA protein production (pro-inflammatory), and enhanced IL10 mRNA expression (anti-inflammatory), leading to suppressed maturation of osteoclast-like cells and enhanced maturation of osteoblasts in vitro. The findings suggest that B. vulgatus and VA may represent promising targets for osteoporosis prevention/treatment.

Background: Osteoporosis is a major global health issue, weakening bones and increasing fracture risk. Dual-energy X-ray absorptiometry (DXA) is the standard for measuring bone mineral density (BMD) and diagnosing osteoporosis, but its costliness and complexity impede widespread screening adoption. Predictive modeling using genetic and clinical data offers a cost-effective alternative for assessing osteoporosis and fracture risk. This study aims to develop BMD prediction models using data from the UK Biobank (UKBB) and test their performance across different ethnic and geographical populations. Methods and findings: We developed BMD prediction models for the femoral neck (FNK) and lumbar spine (SPN) using both genetic variants and clinical factors (such as sex, age, height, and weight), within 17,964 British white individuals from UKBB. Models based on regression with least absolute shrinkage and selection operator (LASSO), selected based on the coefficient of determination (R²) from a model selection subset of 5,973 individuals from British white population. These models were tested on 5 UKBB test sets and 12 independent cohorts of diverse ancestries, totaling over 15,000 individuals. Furthermore, we assessed the correlation of predicted BMDs with fragility fractures risk in 10 years in a case-control set of 287,183 European white participants without DXA-BMDs in the UKBB. With single-nucleotide polymorphism (SNP) inclusion thresholds at 5×10⁻⁶ and 5×10⁻⁷, the prediction models for FNK-BMD and SPN-BMD achieved the highest R² of 27.70% with a 95% confidence interval (CI) of [27.56%, 27.84%] and 48.28% (95% CI [48.23%, 48.34%]), respectively. Adding genetic factors improved predictions slightly, explaining an additional 2.3% variation for FNK-BMD and 3% for SPN-BMD over clinical factors alone. Survival analysis revealed that the predicted FNK-BMD and SPN-BMD were significantly associated with fragility fracture risk in the European white population (P < 0.001). The hazard ratios (HRs) of the predicted FNK-BMD and SPN-BMD were 0.83 (95% CI [0.79, 0.88], corresponding to a 1.44% difference in 10-year absolute risk) and 0.72 (95% CI [0.68, 0.76], corresponding to a 1.64% difference in 10-year absolute risk), respectively, indicating that for every increase of one standard deviation in BMD, the fracture risk will decrease by 17% and 28%, respectively. However, the model’s performance declined in other ethnic groups and independent cohorts. The limitations of this study include differences in clinical factors distribution and the use of only SNPs as genetic factors. Conclusions: In this study, we observed that combining genetic and clinical factors improves BMD prediction compared to clinical factors alone. Adjusting inclusion thresholds for genetic variants (e.g., 5×10⁻⁶ or 5×10⁻⁷) rather than solely considering genome-wide association study (GWAS)-significant variants can enhance the model’s explanatory power. The study highlights the need for training models on diverse populations to improve predictive performance across various ethnic and geographical groups.

Although genome-wide association studies (GWASs) have identified thousands of variants for various traits, the causal variants and the mechanisms underlying the significant loci are largely unknown. In this study, we aim to predict non-coding variants that may functionally affect translation initiation through long-range chromatin interaction. By incorporating the Hi-C data, we propose a novel and powerful deep learning model of artificial intelligence to classify interacting and non-interacting fragment pairs and predict the functional effects of sequence alteration of single nucleotide on chromatin interaction and thus on gene expression. The changes in chromatin interaction probability between the reference sequence and the altered sequence reflect the degree of functional impact for the variant. The model was effective and efficient with the classification of interacting and non-interacting fragment pairs. The predicted causal SNPs that had a larger impact on chromatin interaction were more likely to be identified by GWAS and eQTL analyses. We demonstrate that an integrative approach combining artificial intelligence—deep learning with high throughput experimental evidence of chromatin interaction leads to prioritizing the functional variants in disease- and phenotype-related loci and thus will greatly expedite uncover of the biological mechanism underlying the association identified in genomic studies. Source code used in data preparing and model training is available at the GitHub website (https://github.com/biocai/DeepHiC). Supplementary data are available at Bioinformatics online.

Osteoporosis is a common metabolic bone disease with aging, characterized by low bone mineral density (BMD) and higher fragility fracture risk. Although current pharmacological interventions provide therapeutic benefits, long‐term use is limited by side effects and comorbidities. In this study, we employed driver signaling network identification (DSNI) and drug functional networks (DFN) to identify repurposable drugs from the Library of Integrated Network‐Based Cellular Signatures. We constructed osteoporosis driver signaling networks (ODSN) using multi‐omics data and developed DFN based on drug similarity. By integrating ODSN and DFN with drug‐induced transcriptional responses, we screened 10,158 compounds and identified several drugs with strong targeting effects on ODSN. Mendelian randomization assessed potential causal links between cis‐eQTLs of drug targets and BMD using genome‐wide association study data. Our findings indicate four drugs, including Ruxolitinib, Alfacalcidol, and Doxercalciferol, may exert anti‐osteoporosis effects. Notably, Acebutolol, a β‐blocker for hypertension, has not previously been implicated in osteoporosis therapy. For validation, zebrafish osteoporosis models were established using Dexamethasone‐induced bone loss, followed by treatment with Acebutolol hydrochloride and Alfacalcidol. Both compounds demonstrated significant protective effects against osteoporosis‐related bone deterioration. Furthermore, a population‐based data set, utilizing propensity score matching and analyzed via a generalized linear model, revealed that individuals taking β‐blocker drugs exhibited significantly higher BMD than users of other cardiovascular medications. In summary, this study integrates multi‐omics approaches, experimental validation, and real‐world population data to propose acebutolol as a novel candidate for osteoporosis treatment. These findings warrant further mechanistic studies and clinical trials to evaluate its efficacy in osteoporosis management.

Although gut microbiota influences osteoporosis risk, the individual species involved, and underlying mechanisms, are unknown. We performed integrative analyses in a Chinese cohort with metagenomics/targeted metabolomics/whole-genome sequencing. Bacteroides vulgatus was found negatively associated with bone mineral density (BMD), this association was validated in US Caucasians. Serum valeric acid was positively associated with BMD, and B.vulgatus causally downregulated it. Ovariectomized mice fed B.vulgatus had decreased bone formation and increased bone resorption, lower BMD and poorer bone micro-structure. Valeric acid suppressed NF-κB p65 protein production (pro-inflammatory), and enhanced IL-10 mRNA expression (anti-inflammatory), leading to suppressed maturation of osteoclast-like cells, and enhanced maturation of osteoblasts in vitro. B.vulgatus and valeric acid represent promising targets for osteoporosis prevention/treatment.