Author(s) :

Gan Chen.

Volume/Issue :

Abstract :

Public perceptions of schizophrenia are often shaped by digital narratives that emphasize certain symptoms while minimizing others. This study examines the divergence between clinical symptom severity and online perception using two datasets: one from a clinical PANSS-based schizophrenia assessment and another derived from Reddit posts classified via zero-shot learning. Eleven overlapping symptoms were extracted and scored on a 1–7 scale. Methods included comparative boxplots, Chi-square tests, correlation heatmaps, random forest classification, and logistic regression. The results revealed consistent statistical and perceptual mismatches across domains. Symptoms such as tension and suspiciousness were overrepresented in Reddit posts, while cognitive and affective symptoms like guilt and poor attention were underrepresented. The f indings underscore the need for public education efforts tailored to underrecognized symptoms, and highlight the potential of machine learning in digital psychiatry. Bridging these gaps may improve stigma reduction and enhance digital mental health literacy.