Drug Repurposing Prediction for Immune-Mediated Cutaneous Diseases using a Word-Embedding–Based Machine Learning Approach

Abstract

Immune-mediated diseases affect more than 20% of the population, and many autoimmune diseases affect the skin. Drug repurposing (or repositioning) is a cost-effective approach for finding drugs that can be used to treat diseases for which they are currently not prescribed. We implemented an efficient bioinformatics approach using word embedding to summarize drug information from more than 20 million articles and applied machine learning to model the drug-disease relationship. We trained our drug repurposing approach separately on nine cutaneous diseases (including psoriasis, atopic dermatitis, and alopecia areata) and eight other immunemediated diseases and obtained a mean area under the receiver operating characteristic of 0.93 in crossvalidation. Focusing in particular on psoriasis, a chronic inflammatory condition of skin that affects more than 100 million people worldwide, we were able to confirm drugs that are known to be effective for psoriasis and to identify potential candidates used to treat other diseases. Furthermore, the targets of drug candidates predicted by our approach were significantly enriched among genes differentially expressed in psoriatic lesional skin from a large-scale RNA sequencing cohort. Although our algorithm cannot be used to determine clinical efficacy, our work provides an approach for suggesting drugs for repurposing to immune-mediated cutaneous diseases.

SKiM - A generalized literature-based discovery system for uncovering novel biomedical knowledge from PubMed

“…We filtered the expert curated drugs annotated for ‘therapeutic use’ from CTD. We used the disease-drug association list compiled recently from CTD and NDF-RT 63 and extended it with the disease-drug association from DrugBank and ClinicalKey. DrugBank includes the disease-drug association in the indications section.…”

Machine Learning in Dermatology: Current Applications, Opportunities, and Limitations

“…The best model of these authors predicted treatment outcomes with 18% classification error, demonstrating the utility of basic clinical information. In addition to identifying the treatment response for psoriasis, ML models can also discover potential off-label treatments for psoriasis, atopic dermatitis, and alopecia areata [97]. This model uses a combination of an unsupervised word embedding model summarized drug information from over 20 million articles and application of classification of disease ML models to identify potential drugs for immune-mediated cutaneous diseases.…”

Drug Repurposing in Dermatology: Molecular Biology and Omics Approach

“…Also, Patrick et al gathered drug-related information from more than 20 million articles using machine learning based on word embedding to build a model that highlights drug-disease relationship in order to repurpose drugs for treatment of immune-mediated dermatological conditions, where prednisone (21), triamcinolone (22), budesonide (23), hydroxychloroquine (24), and leflunomide (25) were among the top five predicted drugs for treatment of psoriasis [33]. The chemical structure of compounds (15-25) is demonstrated in Figure 7.…”