Early diagnosis of pulmonary tuberculosis (PTB) is pivotal for achieving effective tuberculosis (TB) control. This study aimed to assess the effectiveness of nanopore sequencing of sputum, bronchoalveolar lavage fluid (BALF), and pleural fluid samples for achieving early PTB diagnosis and provided head-to-head comparisons of nanopore sequencing results versus results obtained using smear, culture, and Xpert MTB/RIF assays. Patients admitted from October 2021 to April 2023 were screened for PTB using diagnostic imaging and electronic medical records. A total of 172 patients (129 PTB, 43 non-TB patients) were included in the final analysis after the exclusion of patients who did not meet the study’s inclusion criteria. PTB-positive rates were determined for each assay, and then, assay diagnostic efficacies were compared. The positive MTB-detection rates obtained using nanopore sequencing were 86.8% for all samples, 62.3% for BALF, and 84.6% for pleural fluid, all of which were significantly higher than the corresponding rates obtained using the other three assays. The overall sensitivity rates, specificity rates, and area under the curve (AUC) values obtained from smear testing were 5.4%, 95.3%, and 0.504, respectively, as compared to the respective results obtained via culture (18.6%, 100.0%, and 0.593), Xpert MTB/RIF (26.4%, 97.7%, and 0.620), and nanopore sequencing (85.3%, 95.4%, and 0.903). The diagnostic efficacy of nanopore sequencing surpassed the diagnostic efficacies of smear, culture, and Xpert MTB/RIF assays. Thus, nanopore sequencing holds promise as an alternative to Xpert MTB/RIF for early PTB detection, particularly for the testing of BALF and pleural fluid samples.

Accurate prediction of Homologous Recombination Deficiency (HRD) is vital for personalized cancer therapy, yet genomic assays are often costly and complex. Predicting HRD from Hematoxylin and Eosin (H&E) stained whole-slide images (WSIs) via deep learning offers an alternative, but model generalization across diverse cancer types remains challenging. We developed a three-stage transfer learning framework to improve HRD prediction across cancer types. After establishing baseline models on data from eight cancer types, the TCGA-BRCA model was selected as a pretrained model for transfer learning applied to eligible cohorts. Transfer learning enhanced HRD prediction performance specifically in cancers with histological similarity to the TCGA-BRCA source model. The AUC increased by + 4% (TCGA-LUAD), + 5% (CPTAC-LUAD), and + 1% (SARC) compared to baseline models. Interpretability analysis confirmed that the model’s predictions were driven by histologically relevant tumor regions, which was further supported by quantitative cellular analysis showing that HRD-high status was characterized by a significantly higher density of neoplastic cells (HRD-high: median 16.0, IQR 1.0–27.0 vs. HRD-low: median 1.0, IQR:0.0–5.0; p < 0.001), whereas HRD-low cases featured a stromal-rich microenvironment. Conclusions This study demonstrates that transfer learning can enhance HRD prediction from H&E images for cancer types sharing histopathological features with the TCGA-BRCA source model, offering a more efficient and accessible approach for clinical HRD assessment.