The human oral microbiome has been known to show strong association with various oral diseases including oral cancer. This study attempts to characterize the community variations between normal, oral potentially malignant disorders (OPMD) and cancer associated microbiota using 16S rDNA sequencing. Swab samples were collected from three groups (normal, OPMD and oral cancer) with nine subjects from each group. Bacteria genomic DNA was isolated in which full length 16S rDNA were amplified and used for cloned library sequencing. 16S rDNA sequences were processed and analysed with MOTHUR. A core oral microbiome was identified consisting of Firmicutes, Proteobacteria, Fusobacteria, Bacteroidetes and Actinobacteria at the phylum level while Streptococcus, Veillonella, Gemella, Granulicatella, Neisseria, Haemophilus, Selenomonas, Fusobacterium, Leptotrichia, Prevotella, Porphyromonas and Lachnoanaerobaculum were detected at the genus level. Firmicutes and Streptococcus were the predominant phylum and genus respectively. Potential oral microbiome memberships unique to normal, OPMD and oral cancer oral cavities were also identified. Analysis of Molecular Variance (AMOVA) showed a significant difference between the normal and the cancer associated oral microbiota but not between the OPMD and the other two groups. However, 2D NMDS showed an overlapping of the OPMD associated oral microbiome between the normal and cancer groups. These findings indicated that oral microbes could be potential biomarkers to distinguish between normal, OPMD and cancer subjects.
Primary diabetes care and diabetic retinopathy (DR) screening persist as major public health challenges due to a shortage of trained primary care physicians (PCPs), particularly in low-resource settings. Here, to bridge the gaps, we developed an integrated image-language system (DeepDR-LLM), combining a large language model (LLM module) and image-based deep learning (DeepDR-Transformer), to provide individualized diabetes management recommendations to PCPs. In a retrospective evaluation, the LLM module demonstrated comparable performance to PCPs and endocrinology residents when tested in English and outperformed PCPs and had comparable performance to endocrinology residents in Chinese. For identifying referable DR, the average PCP's accuracy was 81.0% unassisted and 92.3% assisted by DeepDR-Transformer. Furthermore, we performed a single-center real-world prospective study, deploying DeepDR-LLM. We compared diabetes management adherence of patients under the unassisted PCP arm (n = 397) with those under the PCP+DeepDR-LLM arm (n = 372). Patients with newly diagnosed diabetes in the PCP+DeepDR-LLM arm showed better self-management behaviors throughout follow-up (P