Comprehensive neurodevelopmental assessment through non-specialists: Validation of the STREAM digital platform in India and Malawi

BackgroundChildren in low-resource settings often lack access to culturally appropriate and feasible neurodevelopmental assessments. Existing tools are costly, require specialist training, and were developed in high-resource settings, limiting their scalability. The Scalable TRansdiagnostic Early Assessment of Mental health (STREAM) project uses a mobile platform, delivered by non-specialist workers, to assess motor, social, and cognitive development in children aged 0-6 years. We report STREAM scores, psychometric properties, and age-standardised reference curves from India and Malawi. MethodsWe recruited children aged 0-6 years in New Delhi (India) and Blantyre (Malawi) from two samples: community (clinics and early education centres) and enriched (children with existing/suspected neurodevelopmental conditions). Trained non-specialists administered STREAM in community settings. Motor, cognitive, and social domain scores were derived from age-adjusted data using structural equation modelling (SEM). Validity was assessed against the Griffiths Mental Development Scales (GMDS), caregiver-reported developmental measures, and anthropometric indices. We examined known-groups discriminability, test-retest reliability, and 18-month responsiveness to change. FindingsWe assessed 3992 children (community=3733; enriched=259). SEM fit exceeded standard benchmarks (RMSEA=0.03; CFI=0.98; TLI=0.97). Across both sites, we found extensive evidence for STREAMs criterion validity [range: r=0.31,0.65], convergent validity (9/12 related constructs correlated significantly), and known-groups validity (all hypothesised groups differed significantly). Test-retest reliability was largely good (>0.60&<0.74), and the Indian sample showed some evidence of responsiveness to change. InterpretationWe provide evidence that STREAM is a valid and reliable mobile assessment of neurodevelopment in low-resource settings, designed for delivery by non-specialists in community settings. It captures key neurodevelopmental domains, demonstrates robust psychometric properties, enables standardised monitoring via reference curves, and provides a scalable approach to reducing inequities in access to neurodevelopmental assessments in settings where specialist-delivered tools may not be feasible due to cost, training requirements, or cultural applicability. FundingMedical Research Council (MRC) Global Challenges Research Fund (MR/S036423/1).