Exploring Symptom Connectivity in Colorectal Cancer: A Graph-Theoretical Analysis of Online Patient Discourse

This dataset provides a structured representation of patient-generated narratives extracted from the Reddit community r/coloncancer. The primary objective of this collection was to capture spontaneous, unprompted discourse related to the lived experience of colon cancer outside of traditional clinical environments.Data Collection and Composition:The corpus comprises 76,510 paragraph-level narrative units derived from public posts and comments.Totaling 422,486 raw tokens, the text was processed in its original English form to preserve natural linguistic nuances.A specific subset focusing on depression-related language was identified using predefined affective lexical criteria to facilitate comparative structural analysis.Processing and Methodological Framework:The raw textual data was transformed into lexical co-occurrence networks using a sliding window approach (window size = 5).To ensure clinical relevance, the narrative tokens were mapped onto a predefined symptom vocabulary derived from the EORTC QLQ-C30 and QLQ-CR29 quality-of-life instruments.The dataset includes the results of a triad-based extraction process, identifying fully connected three-node motifs (triangles) to characterize higher-order relationships between symptoms.This repository includes the de-identified textual corpus and the computational scripts (Python/NetworkX) used for network construction and triad enumeration.