OcIMIDo

OcIMIDo provides an ontology for ocular immune-mediated inflammatory diseases that structures concepts and terminology to enable extraction and analysis of clinical and patient-reported information from unstructured text sources such as online patient forums.

Key Features:

• Comprehensive Class Structure: OcIMIDo v1.2 contains 661 classes describing anatomy, clinical phenotypes, disease activity status, complications, investigations, interventions, and functional impacts.
• Rich Relationships and Annotations: The ontology includes 1661 relationships and axioms, 2851 annotations, and 1131 database cross-references linking terms to external biomedical resources.
• Inclusion of Patient-Preferred Synonyms: It incorporates 187 patient-preferred synonyms text-mined from oliviasvision.org, added using a statistical ranking approach and validated by split-sampling against manual extraction.
• Sentiment Analysis Capability: OcIMIDo has been applied to sentiment analysis of 9031 forum posts (median sentiment 0.12, IQR 0.01–0.24), with multivariable logistic regression showing negative sentiment associated with first posts versus replies (OR 3.3, 95% CI 2.8–3.9, p < 0.001).

Scientific Applications:

• Unstructured Text Analysis: Enables systematic extraction and structuring of information from unstructured patient- or clinician-reported text.
• Phenotype and Treatment Mining: Facilitates identification of frequently mentioned clinical phenotypes, treatments, and complications within patient discussions.
• Real-World Evidence Generation: Supports assessment of disease patterns, treatment efficacy, and patient concerns in real-world forum data.

Methodology:

Development incorporated cross-references to other biomedical ontologies, text-mining of online forums (oliviasvision.org), statistical ranking for synonym selection, split-sampling validation versus manual extraction, and application of sentiment analysis with multivariable logistic regression on forum posts.