Improved Autism Spectrum Disorder Detection Using Euclidean-Distance Facial Landmark Features to Enhance Classification Accuracy and Stability with Cross-Validation
Downloads
Autism Spectrum Disorder (ASD) is a neurodevelopmental condition that affects communication skills, social interaction, and behavioral patterns. Early detection is essential for timely intervention; however, conventional diagnostic methods remain time-consuming and subjective, as they rely heavily on clinical observations and expert judgment. This limitation highlights the need for an automated and objective approach to support early ASD screening. This study aims to analyze the performance, stability, and generalization of ASD classification using geometric features extracted from distances between facial landmarks. By representing facial morphology in terms of quantitative spatial relationships, this approach provides a more interpretable alternative to raw image-based methods. This study contributes by proposing a geometric feature representation based on facial landmark distances, providing a comparative analysis between linear and nonlinear classifiers, and ensuring robust evaluation through cross-validation. The dataset consists of 2,032 facial images, evenly distributed between children with ASD and those with typical development. A total of 68 facial landmark points were detected and used to compute pairwise Euclidean distances as classification features. Two classification algorithms, Logistic Regression and Extra Trees Classifier, were evaluated using 5-fold cross-validation to ensure reliable and unbiased performance estimation. The results show that Logistic Regression achieved an average accuracy of 89.91%, precision of 91.04%, recall of 88.56%, and F1-score of 89.76%. Meanwhile, the Extra Trees Classifier outperformed the linear model, achieving an average accuracy of 91.88%, precision of 92.69%, recall of 90.89%, and F1-score of 91.77%. Overall, both models demonstrated stable and consistent performance across validation folds, with the Extra Trees Classifier showing superior ability to capture nonlinear patterns in the data. These findings indicate that geometric feature extraction based on facial landmark distances is effective for ASD detection and has strong potential to be developed as an objective, interpretable, and efficient early screening tool using children’s facial images.
[1] L. Qin, H. Wang, W. Ning, M. Cui, and Q. Wang, “New advances in the diagnosis and treatment of autism spectrum disorders,” European Journal of Medical Research, pp. 1–11, 2024, doi: 10.1186/s40001-024-01916-2.
[2] J. Ostrowski, U. Religioni, B. Gellert, J. Sytnik-Czetwertyński, and J. Pinkas, “Autism Spectrum Disorders: Etiology, Epidemiology, and Challenges for Public Health,” Medical Science Monitor, vol. 30, pp. 1–6, 2024, doi: 10.12659/MSM.944161.
[3] D. Control, “Prevalence and Characteristics of Autism Spectrum Disorder Among Children Aged 8 Years — Autism and Developmental Disabilities Monitoring Network , 11 Sites , United States , 2020,” Morbidity and Mortality Weekly Report, vol. 72, no. 2, 2023, doi: 10.15585/mmwr.ss7202a1.
[4] S. P. Collins et al., “A Study on Autism Detection from Facial Images,” International Journal for Research in Applied Science & Engineering Technology (IJRASET), vol. 11, no. XI, pp. 167–186, 2023, doi: https://doi.org/10.22214/ijraset.2023.56908.
[5] H. M. Dubrock, H. D. Germack, M. Gauthier, L. Jörg, and L. Ambika, “Economic Burden of Delayed Diagnosis in Patients with Pulmonary Arterial Hypertension ( PAH ),” PharmacoEconomics - Open, vol. 8, no. 1, pp. 133–146, 2024, doi: 10.1007/s41669-023-00453-8.
[6] L. Berbel, A. María, A. Joan, C. Marta, and A. Annika, “Association Between Diagnostic Delay and Economic and Clinical Burden in Axial Spondyloarthritis : A Multicentre Retrospective Observational Study,” Rheumatology and Therapy, vol. 12, no. 2, pp. 255–266, 2025, doi: 10.1007/s40744-024-00742-6.
[7] P. McCarty and R. E. Frye, “Early Detection and Diagnosis of Autism Spectrum Disorder: Why Is It So Difficult?,” Seminars in Pediatric Neurology, vol. 35, 2020, doi: 10.1016/j.spen.2020.100831.
[8] M. A. Mahmood, L. Jamel, N. Alturki, and M. A. Tawfeek, “Leveraging artificial intelligence for diagnosis of children autism through facial expressions,” Scientific Reports, vol. 15, no. 1, pp. 1–20, 2025, doi: 10.1038/s41598-025-96014-6.
[9] G. Quatrosi et al., “Cranio-Facial Characteristics in Autism Spectrum Disorder: A Scoping Review,” Journal of Clinical Medicine, vol. 13, no. 3, 2024, doi: 10.3390/jcm13030729.
[10] M. S. ALAM, M. M. RASHID, A. R. FAIZABADI, and H. F. M. ZAKI, “Power of Alignment: Exploring the Effect of Face Alignment on Asd Diagnosis Using Facial Images,” IIUM Engineering Journal, vol. 25, no. 1, pp. 317–327, 2024, doi: 10.31436/iiumej.v25i1.2838.
[11] K. K. Mujeeb Rahman and M. Monica Subashini, “Identification of Autism in Children Using Static Facial Features and Deep Neural Networks,” Brain Sci, vol. 12, 2022, doi: 10.3390/brainsci12010094.
[12] M. Faheem, I. Ahmad, J. Rashid, N. A. Khan, and R. Amin, “Autism spectrum disorder detection using facial images : A performance comparison of pretrained convolutional neural networks,” Healthcare Technology Letters published, vol. 11, pp. 227–239, 2024, doi: 10.1049/htl2.12073.
[13] P. Reddy and J. Andrew, “Diagnosis of Autism in Children Using Deep Learning Techniques by Analyzing Facial Features †,” vol. 59, no. 1, p. 198, 2024, doi: https://doi.org/10.3390/engproc2023059198.
[14] M. F. Almufareh, S. Tehsin, and M. Humayun, “Facial Classification for Autism Spectrum Disorder,” vol. 3, pp. 1–8, 2024, doi: 10.57197/JDR-2024-0025.
[15] T. Farhat, S. Akram, H. S. Alsagri, Z. Ali, A. Ahmad, and A. Jaffar, “Facial Image-Based Autism Detection : A Comparative Study of Deep Neural Network Classifiers,” Computers, Materials & Continua, vol. 78, no. 1, pp. 105–126, 2024, doi: 10.32604/cmc.2023.045022.
[16] F. Mohammadi, “Artificial Intelligence in Autism Spectrum Disorder Diagnosis : A Scoping Review of Face , Voice , and Text Analysis Methods,” Health Science Reports, vol. 8, no. 11, 2025, doi: 10.1002/hsr2.71476.
[17] M. Melinda, M. Oktiana, Y. Nurdin, I. Pujiati, and M. Irhamsyah, “Performance of ShuffleNet and VGG-19 Architectural Classification Models for Face Recognition in Autistic Children,” International Journal on Advance Science Engineering Information Technology, vol. 13, no. 2, pp. 674–680, 2023.
[18] S. Yoonesi, R. A. Azar, M. A. Bafrani, S. Yaghmayee, and H. Shahavand, “Facial expression deep learning algorithms in the detection of neurological disorders : a systematic review and meta analysis,” BioMedical Engineering OnLine, vol. 24, 2025, doi: 10.1186/s12938-025-01396-3.
[19] Q. Teng, Z. Liu, Y. Song, K. Han, and Y. Lu, “A survey on the interpretability of deep learning in medical diagnosis,” Multimedia Systems, vol. 28, pp. 2335–2355, 2022, doi: 10.1007/s00530-022-00960-4.
[20] B. Banire, D. Al Thani, M. Qaraqe, and B. Mansoor, “Face-Based Attention Recognition Model for Children with Autism Spectrum Disorder,” Journal of Healthcare Informatics Research, vol. 5, no. 4, pp. 420–445, 2021, doi: 10.1007/s41666-021-00101-y.
[21] B. Al-baker, X. Ju, P. Mossey, and A. Ayoub, “The accuracy of automated facial landmarking - a comparative study between Cliniface software and patch- based Convoluted Neural Network algorithm,” European Journal of Orthodontics, vol. 47, 2025, doi: 10.1093/ejo/cjaf009.
[22] K. Rezaee, T. S. Javanmardi, and A. M. Heydari, “A deep learning framework for classifying autism spectrum disorder from children ’ s facial images using a multi-scale ViT architecture and edge computing,” Exploration of Medicine, vol. 6, 2025, doi: 10.37349/emed.2025.1001376.
[23] G. Sanil, K. Prakash, S. Prabhu, V. C. Nayak, and S. Sengupta, “2D-3D Facial Image Analysis for Identification of Facial Features Using Machine Learning Algorithms With Hyper-Parameter Optimization for Forensics Applications,” IEEE Access, vol. 11, no. August, pp. 82521–82538, 2023, doi: 10.1109/ACCESS.2023.3298443.
[24] Y. An, J. Lee, E. Bak, and S. Pan, “Deep Facial Emotion Recognition Using Local Features Based on Facial Landmarks for Security System,” Computers, Materianls and Continua, vol. 76, p. 2, 2023, doi: 10.32604/cmc.2023.039460.
[25] Q. Y. Shambour, N. Qandeel, Y. Alraba’nah, A. Abumariam, and M. K. Shambour, “Artificial Intelligence Techniques for Early Autism Detection in Toddlers: A Comparative Analysis,” Journal of Applied Data Sciences, vol. 5, no. 4, pp. 1754–1764, 2024, doi: 10.47738/jads.v5i4.353.
[26] L. Kampa, K. Yamini, A. Basavaraju, and K. Anoop, “A Stack Based Ensemble Learning Method for Diagnosing Autism Spectrum Disorder,” Mathematical Statistician and Engineering Applications, vol. 71, no. 3, pp. 237–251, 2022, doi: https://doi.org/10.17762/msea.v71i3.162.
[27] B. Twala and E. Molloy, “On effectively predicting autism spectrum disorder therapy using an ensemble of classifiers,” Scientific Reports, vol. 13, no. 1, pp. 1–18, 2023, doi: 10.1038/s41598-023-46379-3.
[28] T. J. Bradshaw, Zachary Huemann, Junjie Hu, and Arman Rahmim, “A Guide to Cross-Validation for Artificial Intelligence in,” Radiology: Artificial Intelligence, vol. 5, no. 12, 2023, doi: 10.1148/ryai.220232.
[29] K. Takahashi, Y. Shimamura, C. Tachiki, Y. Nishii, and M. Hagiwara, “Cephalometric landmark detection without X-rays combining coordinate regression and heatmap regression,” Scientific Reports, vol. 13, no. 1, pp. 1–13, 2023, doi: 10.1038/s41598-023-46919-x.
[30] T. Terada, R. Kimura, and Y. Chen, “3D facial landmark detection based on differential cylindrical projection and multi-task learning,” Communications in Information and Systems, vol. 20, no. 4, pp. 443–459, 2020, doi: 10.4310/CIS.2020.v20.n4.a3.
[31] R. S. Raj, D. Pratiba, and R. P. Kumar, “Facial Expression Recognition using Facial Landmarks: A novel approach,” Advances in Science, Technology and Engineering Systems, vol. 5, no. 5, pp. 24–28, 2020, doi: 10.25046/aj050504.
[32] W. Yu-ching, J. Jhen-yu, and T. Lan, “The Accuracy of Facial Landmark Models in Different Facial Expressions,” Journal of Clinical & Experimental Dermatology Research, vol. 15, pp. 3–6, 2024, doi: 10.35841/2155-9554.24.15.674.Citation.
[33] G. Gao, C. Chen, K. Xu, K. Liu, and A. Mashhadi, “Automatic face detection based on bidirectional recurrent neural network optimized by improved Ebola optimization search algorithm,” Scientific Reports, vol. 14, pp. 1–19, 2024, doi: 10.1038/s41598-024-79067-x 1.
[34] R. H. Al-abboodi, “Facial Expression Recognition Based on GSO Enhanced Deep Learning in IOT Environment,” International Journal of Intelligent Engineering and Systems, vol. 17, no. 3, pp. 445–459, 2024, doi: 10.22266/ijies2024.0630.35.
[35] M. A. Khan, A. Power, P. Corcoran, and C. Thorpe, “An Evaluation of Features Extracted from Facial Images in the Context of Binary Age Classification,” Learning and Analytics in Intelligent Systems, vol. 48, pp. 99–112, 2025, doi: 10.1007/978-3-031-88649-2_11.
[36] R. Hidayat, M. O. B. Wibowo, B. Y. Satria, and A. Winursito, “Implementation of Face Recognition Using Geometric Features Extraction,” Jurnal Ilmiah Kursor, vol. 11, no. 2, p. 83, 2022, doi: 10.21107/kursor.v11i2.284.
[37] S. Li, J. Wang, L. Tian, J. Wang, and Y. Huang, “A fine-grained human facial key feature extraction and fusion method for emotion recognition,” Scientific Reports, vol. 15, no. 1, pp. 1–29, 2025, doi: 10.1038/s41598-025-90440-2.
[38] J. J. Levy, J. J. Levy, J. J. Levy, and A. J. O’Malley, “Don’t dismiss logistic regression: The case for sensible extraction of interactions in the era of machine learning,” BMC Medical Research Methodology, vol. 20, no. 1, pp. 1–15, 2020, doi: 10.1186/s12874-020-01046-3.
[39] M. Anshori, M. S. Haris, and A. Wahyudi, “Logistic Regression’s Effectiveness in Feature Selection with Information Gain in Predicting Heart Failure Patients,” Journal of Enhanced Studies in Informatics and Computer Applications, vol. 1, no. 2, pp. 35–39, 2024, doi: 10.47794/jesica.v1i2.8.
[40] O. Old, “Benefits and challenges of using logistic regression to assess neuropsychological performance validity: evidence from a simulation study,” Clinical Neuropsychology, vol. 37, no. 1, pp. 34–59, 2023, doi: 10.1007/978-3-658-38618-4_5.
[41] E. Dumitrescu, S. Hué, C. Hurlin, and S. Tokpavi, “Machine learning for credit scoring: Improving logistic regression with non-linear decision-tree effects,” European Journal of Operational Research, vol. 297, no. 3, pp. 1178–1192, 2022, doi: 10.1016/j.ejor.2021.06.053.
[42] T. E. Mathew, “an Optimized Extremely Randomized Tree Model for Breast Cancer Classification,” Journal of Theoretical and Applied Information Technology, vol. 100, no. 16, pp. 5234–5246, 2022.
[43] F. G. Arage, Z. B. Tadese, E. A. Taye, T. K. Tsegaw, T. G. Abate, and E. A. Alemu, “Cervical cancer screening uptake and its associated factor in Sub-Sharan Africa: a machine learning approach,” BMC Medical Informatics and Decision Making, vol. 25, no. 1, 2025, doi: 10.1186/s12911-025-03039-y.
[44] Y. I. Sulistya and M. Istighosah, “Obesity Prediction with Machine Learning Models Comparing Various Algorithm Performances,” International Journal of Artificial Intelligence in Medical Issues, vol. 3, no. 1, pp. 1–13, 2025, doi: 10.56705/ijaimi.v3i1.181.
[45] J. Li, “Asymptotics of K-Fold Cross Validation,” Journal of Artificial Intelligence Research, vol. 78, pp. 491–526, 2023, doi: https://doi.org/10.1613/jair.1.13974.
[46] V. Singh, M. Pencina, A. J. Einstein, J. X. Liang, D. S. Berman, and P. Slomka, “Impact of train / test sample regimen on performance estimate stability of machine learning in cardiovascular imaging,” Scientific Reports, vol. 11, pp. 1–8, 2021, doi: 10.1038/s41598-021-93651-5.
[47] J. Qiu, “An Analysis of Model Evaluation with Cross-Validation : Techniques , Applications , and Recent Advances,” vol. 0, pp. 69–72, 2024, doi: 10.54254/2754-1169/99/2024OX0213.
[48] V. Teodorescu and L. O. Bras, “Assessing the Validity of k-Fold Cross-Validation for Model Selection : Evidence from Bankruptcy Prediction Using Random Forest and XGBoost,” vol. 13, p. 127, 2025, doi: 10.3390/computation13050127.
[49] J. White and S. D. Power, “k-Fold Cross-Validation Can Significantly Over-Estimate True Classification Accuracy in Common EEG-Based Passive BCI Experimental Designs : An Empirical Investigation,” vol. 23, p. 6077, 2023, doi: 10.3390/s23136077.
[50] Y. Wang, M. Khodadadzadeh, and R. Zurita-milla, “Spatial + : A new cross-validation method to evaluate geospatial machine learning models,” International Journal of Applied Earth Observation and Geoinformation, vol. 121, p. 103364, 2023, doi: 10.1016/j.jag.2023.103364.
[51] T. Abedin, H. Xu, and S. Uddin, “The impact of K selection in K fold cross-validation on bias and variance in supervised learning models,” vol. 16, pp. 1–13, 2026, doi: 10.1038/s41598-026-37247-x 1.
[52] J. C. Obi, “A Comparative Study of Several Classification Metrics and Their Performances on Data,” World Journal of Advanced Engineering Technology and Sciences, vol. 8, no. 1, pp. 308–314, 2023, doi: 10.30574/wjaets.2023.8.1.0054.
[53] D. Chicco and G. Jurman, “An Invitation to Greater Use of Matthews Correlation Coef fi cient in Robotics and Arti fi cial Intelligence,” Frontiers in Robotics and AI, vol. 9, pp. 1–4, 2022, doi: 10.3389/frobt.2022.876814.
[54] G. M. F. Id, “Challenges in the real world use of classification accuracy metrics : From recall and precision to the Matthews correlation coefficient,” PLoS ONE, vol. 18, no. 10, pp. 1–27, 2023, doi: 10.1371/journal.pone.0291908.
Copyright (c) 2026 Yudha Nurdin, Syifa Anzella, Melinda, Junidar, Rosminazuin Ab Rahim (Author)
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution-ShareAlikel 4.0 International (CC BY-SA 4.0) that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
