Autism Spectrum Disorder (ASD) Risk Prediction Using Adaptive Risk Fusion Network with Deep Neural Network (ARFN-DNN)

There is a need for computational screening techniques to supplement traditional assessment since delayed diagnosis of Autism Spectrum Disorder (ASD) can limit the possibility for prompt developmental support. This study suggests an Adaptive likelihood Fusion Network with Deep Neural Network (ARFN-DNN) that uses clinical-profile features and structured screening to predict the likelihood of ASD in children. Through probability-level fusion, the suggested architecture integrates Random Forest (RF), XGBoost (XGB), Logistic Regression (LR), and a regularized Deep Neural Network (DNN).. In order to illustrate behavioral, developmental, clinical, social-communication, and behavioral-communication link the dataset is printed six domain-inspired risk features. For the neural meta-learner, unbiased base-model predictions were produced using a five-fold stratified out-of-fold process. The suggested ARFN-DNN obtained 91.88% accuracy, 91.67% precision, 89.53% recall, 90.59% F1-score, 93.69% specificity, and 0.9424 ROC-AUC on the frozen unseen test set with 197 samples. With a train-test accuracy gap of just 1.89%, the model's generalization was comparatively steady. The results imply that while formal clinical evaluation and external validation are still necessary, probability-level fusion can offer a helpful additional screening layer.

Autism Spectrum Disorder, ASD Screening, Machine Learning, Deep Neural Network, Random Forest, XGBoost, Logistic Regression, SMOTE, Ensemble Learning, Risk Prediction, ARFN-DNN.

1. L. Breiman, "Random forests," Springer, Dordrecht, Netherlands, 2001. [Google Scholar] [Crossref]

2. N. V. Chawla, K. W. Bowyer, L. O. Hall, and W. P. Kegelmeyer, "SMOTE: Synthetic minority over-sampling technique," J. Artif. Intell. Res., AI Access Foundation, Mountain View, CA, USA, 2002. [Google Scholar] [Crossref]

3. T. Chen and C. Guestrin, "XGBoost: A scalable tree boosting system," Association for Computing Machinery, New York, NY, USA, 2016. [Google Scholar] [Crossref]

4. S. M. Lundberg and S.-I. Lee, "A unified approach to interpreting model predictions," Neural Information Processing Systems Foundation, Long Beach, CA, USA, 2017. [Google Scholar] [Crossref]

5. I. Loshchilov and F. Hutter, "Decoupled weight decay regularization," in Proc. Int. Conf. on Learning Representations, New Orleans, LA, USA, 2019. [Google Scholar] [Crossref]

6. D. Pavithra, A. N. Jayanthi, R. Nidhya, and S. Balamurugan, "Autism screening tools with machine learning and deep learning methods: A review," Wiley, Hoboken, NJ, USA, 2022. [Google Scholar] [Crossref]

7. R. Francese and X. Yang, "Supporting autism spectrum disorder screening and intervention with machine learning and wearables: A systematic literature review," Springer Nature, Cham, Switzerland, 2022. [Google Scholar] [Crossref]

8. M. E. Minissi, I. A. Chicchi Giglioli, F. Mantovani et al., "Assessment of the autism spectrum disorder based on machine learning and social visual attention: A systematic review," Springer Nature, Cham, Switzerland, 2022. [Google Scholar] [Crossref]

9. M. R. Lamani and J. B. Pernabas, "A thorough review of deep learning in autism spectrum disorder detection: From data to diagnosis," Bentham Science Publishers, Sharjah, UAE, 2024. [Google Scholar] [Crossref]

10. World Health Organization, "Autism," World Health Organization, Geneva, Switzerland, 2025. [Google Scholar] [Crossref]

11. R. Blitz and S. Swoboda, "How Pediatricians Screen for Autism," American Academy of Pediatrics, Itasca, IL, USA, 2025. [Google Scholar] [Crossref]

12. M. A. Zoromba et al., "Artificial intelligence methods in early detection of autism spectrum disorder: A DSM-5 criterion-based systematic review," Wiley, Hoboken, NJ, USA, 2026. [Google Scholar] [Crossref]

13. Scikit-learn Developers, "LogisticRegression," Scikit-learn Project, Online Documentation, USA, 2026. [Google Scholar] [Crossref]

14. Anupama N. and C. M. Patil, "Data-driven approaches for autism detection: A comprehensive review of machine learning algorithms and datasets," Wiley, Hoboken, NJ, USA, 2026. [Google Scholar] [Crossref]

15. K. Srinivasa Reddy and Y. Lakshmi Annapurna, "A scoping review of machine learning and deep learning methods for autism spectrum disorder diagnosis and analysis," J. Vis. Exp., Cambridge, MA, USA, 2026. [Google Scholar] [Crossref]