AI-BASED DEVELOPMENTAL SCREENING VS. CURRENT SURVEILLANCE: THE ECONOMIC VALUE OF EARLY AUTISM DETECTION IN MALAYSIA
Author(s)
Hui Yee Yeo, PhD, MSc (Health Economics), MPharm (Hons), RPh1, Sivaraj Raman, MSc (Pharmacoeconomics), MPharm (Hons), RPh2, Nor Zam Azihan Mohd Hassan, DrPH, MD2, Ai Ch'i Liew, MSc (Clinical Pharmacy), BPharm (Hons), RPh3, Amar-Singh HSS, MBBS (Mal), MRCP (UK), FRCP (Glasg)4.
1Immunisation Advisory Centre, Gen. Practices & Primary Hlthcare, and Liggins Institute, University of Auckland, Auckland, New Zealand, 2Institute for Health Systems Research, National Institutes of Health, Ministry of Health Malaysia, Selangor, Malaysia, 3Clinical Research Centre, Hospital Seberang Jaya, Ministry of Health Malaysia, Seberang Jaya, Malaysia, 4Galen Centre for Health and Social Policy, Kuala Lumpur, Malaysia.
1Immunisation Advisory Centre, Gen. Practices & Primary Hlthcare, and Liggins Institute, University of Auckland, Auckland, New Zealand, 2Institute for Health Systems Research, National Institutes of Health, Ministry of Health Malaysia, Selangor, Malaysia, 3Clinical Research Centre, Hospital Seberang Jaya, Ministry of Health Malaysia, Seberang Jaya, Malaysia, 4Galen Centre for Health and Social Policy, Kuala Lumpur, Malaysia.
OBJECTIVES: To evaluate the economic value of addressing limitations in Malaysia’s current autism spectrum disorder (ASD) surveillance system through implementation of a nationwide artificial intelligence (AI)-based developmental screening programme.
METHODS: A decision tree model evaluated the economic impact of the screening programme in 174,036 three-year-old Malaysian preschoolers versus existing surveillance across three interconnected modules: screening/diagnosis, intervention, and lifetime economic outcomes. Model inputs included ASD prevalence, sensitivity and specificity of screening/surveillance, intervention timing and intensity, and costs. Economic outcomes encompassed educational needs, productivity (estimated using the human capital approach), and the level of independence. Parameters were informed by the national health dataset, local case-control study, government policies, literature reviews, stakeholders’ consultations, and Delphi expert consensus. Cost-benefit analysis from a societal perspective over a lifetime horizon using a 3% discount rate was conducted to quantify the incremental net monetary benefit (INMB) between strategies. One-way sensitivity analyses assessed parameter uncertainty.
RESULTS: Compared to surveillance, an AI-based nationwide screening program increased early ASD identification by 5.5-fold (1,732 vs 315), enabling timely intervention and improved developmental trajectories toward milder ASD severity. Although screening and diagnostic costs were higher for the screening programme (MYR 35.64 million vs. MYR 7.03 million), total expenditures were reduced by MYR 636.69 million due to lower downstream intervention costs. Additional lifetime societal benefits included MYR 1.37 billion from improved independence, MYR 596.50 million from increased productivity, and MYR 177.55 million from reduced educational support needs, resulting in a total INMB of MYR 1.04 billion. Sensitivity analyses identified screening sensitivity, discount rate, ASD prevalence, cohort size, and diagnostic non-attendance as key drivers of economic values.
CONCLUSIONS: Current ASD surveillance in Malaysia may miss opportunities for early diagnosis and intervention. Nationwide AI-enabled developmental screening could address these gaps and generate substantial long-term societal and economic benefits.
METHODS: A decision tree model evaluated the economic impact of the screening programme in 174,036 three-year-old Malaysian preschoolers versus existing surveillance across three interconnected modules: screening/diagnosis, intervention, and lifetime economic outcomes. Model inputs included ASD prevalence, sensitivity and specificity of screening/surveillance, intervention timing and intensity, and costs. Economic outcomes encompassed educational needs, productivity (estimated using the human capital approach), and the level of independence. Parameters were informed by the national health dataset, local case-control study, government policies, literature reviews, stakeholders’ consultations, and Delphi expert consensus. Cost-benefit analysis from a societal perspective over a lifetime horizon using a 3% discount rate was conducted to quantify the incremental net monetary benefit (INMB) between strategies. One-way sensitivity analyses assessed parameter uncertainty.
RESULTS: Compared to surveillance, an AI-based nationwide screening program increased early ASD identification by 5.5-fold (1,732 vs 315), enabling timely intervention and improved developmental trajectories toward milder ASD severity. Although screening and diagnostic costs were higher for the screening programme (MYR 35.64 million vs. MYR 7.03 million), total expenditures were reduced by MYR 636.69 million due to lower downstream intervention costs. Additional lifetime societal benefits included MYR 1.37 billion from improved independence, MYR 596.50 million from increased productivity, and MYR 177.55 million from reduced educational support needs, resulting in a total INMB of MYR 1.04 billion. Sensitivity analyses identified screening sensitivity, discount rate, ASD prevalence, cohort size, and diagnostic non-attendance as key drivers of economic values.
CONCLUSIONS: Current ASD surveillance in Malaysia may miss opportunities for early diagnosis and intervention. Nationwide AI-enabled developmental screening could address these gaps and generate substantial long-term societal and economic benefits.
Conference/Value in Health Info
2026-09, ISPOR Asia Pacific 2026, Bangkok, Thailand
Value in Health, Volume 55, Issue S1
Code
EE64
Topic
Economic Evaluation
Disease
SDC: Neurological Disorders, SDC: Pediatrics