Purpose: This study aims to design and develop an Android-based expert system for early detection of Autism Spectrum Disorder in children using the Forward Chaining inference method. The system supports parents and educators in recognizing symptoms and bridging the gap between early identification and professional intervention.

Methodology/approach: This research adopts a research and development approach using a prototype model. Data were collected through literature review, observation, and expert interviews with child development specialists. The expert system applies Forward Chaining using a rule-based knowledge base covering three ASD severity levels and 27 validated symptoms. System performance was evaluated.

Results/findings: The study developed an Android-based expert system to identify autism symptoms and classify severity levels. The system achieved 92% accuracy compared with expert diagnoses, while functional testing confirmed all features operated correctly, including symptom input, diagnostic results, and online clinic reservation.

Conclusions: The Android-based expert system using the Forward Chaining method is effective and reliable for supporting early autism detection. Its logical and transparent inference process makes it suitable for non-expert users, while the integration with healthcare services strengthens early intervention efforts.

Limitations: The system relies on predefined rules and symptom data, which may not capture the full variability of autism manifestations. The application also does not replace professional clinical diagnosis and is limited to early screening purposes.

Contribution: This study contributes a lightweight and accessible mobile expert system for autism detection, integrating diagnostic support with professional access, delivering practical value for parents, educators, and early childhood intervention programs.