Discovering causal paths to diabetic nephropathy by combining computable biomedical knowledge with graph mining...

Area of Science:

• Biomedical informatics
• Graph theory
• Computational biology

Background:

• Diabetic nephropathy (DN) is a major complication of diabetes.
• Understanding the causal pathways of DN is crucial for effective treatment.

Purpose of the Study:

• To identify causal paths from diabetes to diabetic nephropathy using SemMedDB and graph mining.
• To develop an efficient approach for discovering causal paths in complex disorders.

Main Methods:

• Utilized 12,662 triples from SemMedDB, encompassing 3,374 unique concepts and 44 semantic relations.
• Constructed a directed knowledge graph (KG) and pruned it to a causal graph using word2vec, semantic relations, and path length.
• Validated identified causal paths and third variables with a nephrologist.

Main Results:

• Identified a key causal path from diabetes to DN, highlighting pathogenesis inducers and clinical outcomes.
• Observed an increase in causal paths and emerging third variables as the directed causal score decreased from Quantile 95% to Quantile 75%.

Conclusions:

• Developed an efficient causal path discovery method for complex diseases.
• The approach successfully elucidated predominant causal pathways from pathogenesis to disease manifestation.