This scoping review aimed to map the available evidence on the use of health databases to support the early identification of autism spectrum disorder (ASD) across diverse populations and healthcare settings. Early identification was defined as detecting features related to ASD prior to a formal clinical diagnosis.
The review was conducted following the Joanna Briggs Institute scoping review framework. Systematic searches were performed in MEDLINE, Embase, Scopus, PsycINFO, Web of Science and Latin American and Caribbean Health Sciences Literature. Grey literature searches included ProQuest Dissertations and Theses. The search timeframe extended up to December 2024 and was updated in March 2026.
Included studies enrolled individuals with diagnosed or suspected ASD without restrictions on age, country, or healthcare setting. The review considered health databases defined as digital clinical or surveillance sources — for example, electronic medical records and automated healthcare data systems — and explicitly excluded databases based solely on biological or genetic data. The scope prioritized detection of ASD-related features before formal diagnostic confirmation.
Study selection and data extraction were undertaken independently by two reviewers. Extracted elements were synthesised descriptively and organised by database type, analytical methods used, and reported outcomes related to early identification, predictors, comorbidities, and healthcare utilisation.
Thirty-seven studies met inclusion criteria. The majority were conducted in high-income countries and predominantly involved children or adolescents. The review reports that only one of the identified databases was publicly accessible. The remaining data sources were not publicly available, limiting external validation and reuse.
Identified data sources were grouped into three main categories:
Only one database among those reviewed was reported as publicly accessible, indicating limited open data availability in this area.
Studies applied a range of analytical approaches to identify early signals of ASD within health databases:
These approaches reflect a mix of structured-data and unstructured-text analytics across included studies.
Common early predictors reported across studies included:
Reported comorbidities captured within health databases included epilepsy, attention-deficit/hyperactivity disorder (ADHD), mood and anxiety disorders, and sensory issues. These comorbid conditions were frequently documented and associated with children who were later diagnosed with ASD.
Studies consistently found that children later diagnosed with ASD had more frequent medical visits, hospitalisations, and emergency department use prior to diagnosis. Despite early parental concerns being recorded in some settings, diagnostic delays were more pronounced among low-income and underrepresented groups, indicating disparities in timely access to diagnostic assessment and services.
The review highlights several barriers that limit clinical application of database-driven early detection:
The review does not report specific solutions but emphasises that these challenges must be addressed to move from research demonstrations to equitable clinical implementation.
Health databases are increasingly used to support early identification of autism spectrum disorder using methods such as machine learning and natural language processing. The evidence base (37 studies) shows potential for detecting early predictors, comorbidities, and patterns of health service use that precede formal ASD diagnosis. However, the clinical application of these approaches is limited by standardisation gaps, ethical and feasibility concerns, and restricted data access. Addressing these barriers is essential to develop inclusive, evidence-based strategies that can be implemented in practice and that reduce diagnostic disparities among underserved populations.
The review underscores a need for efforts to harmonise data standards, clarify ethical frameworks for predictive analytics in developmental conditions, improve data accessibility for validation, and study implementation pathways that link early detection signals to timely assessment and intervention.
The review reports registration with the Open Science Framework and provides a DOI: 10.17605/OSF.IO/RMVWE. Specific protocol details beyond the registration note were not reported in the source summary provided.