Background: The understanding of drug disposition and response in the maternal and pediatric populations is insufficient. Thus, to facilitate future research, we aim to develop an open-source knowledgebase and portal that provides comprehensive information on all pharmacologic and therapeutic studies conducted within these populations. Methods: Four stages were involved - 1) ontology development, 2) text classification using natural language processing (NLP), 3) knowledgebase development, and 4) portal development. Results: First, ontology development started with classification of studies by the two main populations, i.e., maternal, and pediatric. Further classifications included sub-populations, studies, sub-studies, and study components including the pharmacologic elements. Pharmacologic data varies based on the study type, i.e., pharmaco-kinetics, -dynamics, -metrics, -epidemiology, clinical trials, or omics. Second, we developed a preliminary model to help identify drug-based maternal and pediatric articles in an efficient manner. A 5-fold cross-validation of an NLP model using positively and negatively labeled PubMed abstracts yielded an average accuracy of 94.6% and recall of 96.2%. Lastly, the knowledgebase and portal development are ongoing with a) data curation from the model-retrieved PubMed articles using a customized curation tool, and b) web-user interface with query options for drug names, sub-populations, pregnancy stages or study types. Conclusion: The MPRINT knowledgebase and portal will serve as a one-stop resource for maternal and pediatric studies. It will aid in the identification of knowledge gaps and areas of potential inquiry, thus significantly accelerating therapeutics and translational research conducted in these populations.
