P071
TB Portals and the Quest for the Universal Reference Dataset
A Gabrielian(1) G Rosenfeld(1) K Wollenberg(1) D Hurt(1) A Rosenthal(1)
1:National Institute of Allergy and Infectious Diseases
The TB Portals database stands at the forefront of tuberculosis (TB) research, offering an invaluable reference dataset, useful for research, education, and clinical studies.
Broad and Deep coverage of clinical cases: TB Portals encompasses a wide array of annotations, including socioeconomic data, clinical treatment and outcomes, imaging (X-rays and CT scans), and genomic sequences.
Customizable data queries to enhance clinical studies: The TB Portals platform allows for tailored data queries, enabling researchers to extract specific datasets relevant to their study objectives, thereby enhancing study design and hypothesis testing.
Hard-to-find data from multiple countries: The database serves as an essential benchmark for developing and testing ML and AI algorithms in TB diagnosis and treatment. TB Portals offer less biased and more diverse dataset, coming from multiple hospitals in 19 countries with heavy DR-TB burden.
Accelerating drug and vaccine development: Access to TB Portals' comprehensive genomics data may help in creating new TB drugs and vaccines by enabling more nuanced analyses of pathogen’s evolutionary blueprint against the response to variety of drugs, while taking into account efficiencies of existing genotypic and phenotypic DSTs.
We will present our analysis of TB Portals cohorts, specifically selected to highlight importance of comorbidities (HIV, anemia, diabetes), genomic lineages of M. tuberculosis, and incidence of lesions in lungs. With almost 14,000 clinical cases, it is now possible to perform statistical analysis and machine learning, and enhance many clinical studies with TB Portals data.