Advancing Research Activities for Identifying Novel Therapies

The Challenge

NuMedii was looking for a smart solution for analyzing research literature that would facilitate the identification of new therapies for treating idiopathic pulmonary fibrosis (IPF). The required solution had to be able to leverage both structured (from public and proprietary datasets) and unstructured data (from scientific journals).

There were various challenges in achieving this goal:

• Providing provenance for each underlying fact supporting the scientific conclusions
• Rapidly growing number of data and sources with genomic, molecular, and other biomedical data describing diseases and medicinal drugs
• Highly fragmented nature of the data coming from multiple sources
• Frequent occurrences of semantic redundancy (ambiguity)
• Time- and effort-consuming aspects of data integration processes as well as maintaining such knowledge on a large scale
• Efficiently extracting relevant meaning from the acquired knowledge

The Solution

In 2017, as part of the project, NuMedii aimed to build an industry-specific knowledge graph integrating over 20 open and commercial public databases, as well as proprietary datasets, covering genomics, proteomics, metabolomics, diseases, and drugs. 

This involved semantic data modeling to normalize diverse data schemas and instances to major biomedical ontologies. 

The solution also created extensive semantically annotated scientific literature corpora, which, combined with the knowledge graph, enabled advanced text analysis pipelines to identify and infer new biomedical concepts and relationships.

The Impact

NuMedii significantly increased its efficiency in identifying novel therapies and cut time and resources on research activities. 

The solution improved user experience by providing access to highly normalized and semantically interlinked data, facilitating the discovery of knowledge locked in disparate documents, enabling pattern identification, and supporting hypothesis testing for IPF treatment.