The digital epidemiology of neurodiversity discourse on Chinese video platforms: a 35,813-record, four-tier mapping on Douyin and Bilibili (2019-2026)

Background: Recognition of adult ADHD and high-functioning autism has risen sharply and social media are repeatedly implicated as an amplifier, but evidence is mostly from English-language Twitter/Reddit; systematic measurement on Chinese video platforms is absent. Methods: A pre-registered six-cluster keyword matrix retrieved public videos across 21 saturation rounds, de-duplicated on URL to a corpus of 35,813 records (30,883 Bilibili; 4,691 Douyin; 149 academic; 90 web), each assigned to one of four tiers (T1 strong relevance, T2 symptom-feature, T3 celebrity/IP context, T4 ecosystem), with multi-label themes within T1; platform sampling, algorithmic availability and crawl-volume inflation were explicitly modelled. Results: T1 content grew 9.7-fold from 2019 (n=41) to 2025 (n=399), with 1,162 further T1 items through October 2026. The corpus was dominated by celebrity/IP content (T3 n=26,978, 75.3%), then ecosystem (T4 13.2%), strong relevance (T1 9.3%), symptom-only (T2 2.1%). Within T1, themes were symptom/diagnosis/medication 32.5%, identity/destigmatization 28.8%, self-management (harness) 17.2%, celebrity/genius framing 9.4%. Cumulative engagement reached ~160 million likes (Douyin) and 6.75 billion views (Bilibili). Conclusions: Chinese ND video discourse shows a years-long, structurally organized emergence rather than a single viral spike; the dominant transmissible form is celebrity labeling, while strong-relevance content forms a funnel from symptom recognition to identity to self-management. The measured quantity is discourse prevalence/retrievability, not clinical prevalence - the digital-age counterpart of the recording-rate/selection confound in historical psychobiography. Data are public videos, not personal health information.