Scholarlink

<p>Machine Learning (ML) has become integral across diverse domains, revolutionizing day-to-day activities. Its capability to analyze existing data and make predictions has significantly enhanced productivity in various fields. This paper introduces a novel Detection and Recommendation system tailored for autistic children. Leveraging ML techniques, our approach focuses on identifying autism in children aged 18 to 24 months and delivering tailored therapy recommendations to foster essential skill development. The data for therapy recommendations has been collected from therapy training centers and verified by professionals. For autism detection, we employ a Rule-based classifier, utilizing a unique combination of the QCHAT screening tool rules and ML algorithms. The Existing Systems achieved an accuracy of 90% using Random forest, 89% using Logistic regression, 89% using Naive Bayes, and 97.5 using Deep Neural Networks. The Proposed model achieved an accuracy of 100% using Random forest, 97% using Logistic regression,90% using Naive Bayes, and 98.4% using Deep Neural Networks. An increase in accuracy can be achieved using a proposed model that combines screening tool rules with ML algorithms. All existing systems for detection have used ML algorithms and not included screening tool rules which play an important role in the detection process. The Recommendation System is a Monolithic Hybrid model that seamlessly integrates content-based and multi-criteria collaborative filtering. The latter generates cohorts of similar children based on multiple similarity metrics, feeding this information into the content-based filtering component. The output of content-based filtering, in turn, serves as an input to an expert system, which formulates therapy recommendations with prioritization based on the severity of symptoms. This innovative combination enhances the precision and personalization of therapy recommendations for autistic children, marking a significant advancement in the field. This approach surpasses the performance of conventional content-based and collaborative filtering recommender systems, ensuring precise and suitable recommendations with a precision of 80%. This detection and recommendation system helps therapists to accurately detect autism and provide therapies of the highest priority for the wellness of children rather than doing it manually.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Detection and Recommendation System for Autistic Children Using Machine Learning

  • 期刊论文

  • 发表日期：

    2025年12月15日

查看全文
  • Kusumalatha Karre,
  • Y. Ramadevi

摘要

Machine Learning (ML) has become integral across diverse domains, revolutionizing day-to-day activities. Its capability to analyze existing data and make predictions has significantly enhanced productivity in various fields. This paper introduces a novel Detection and Recommendation system tailored for autistic children. Leveraging ML techniques, our approach focuses on identifying autism in children aged 18 to 24 months and delivering tailored therapy recommendations to foster essential skill development. The data for therapy recommendations has been collected from therapy training centers and verified by professionals. For autism detection, we employ a Rule-based classifier, utilizing a unique combination of the QCHAT screening tool rules and ML algorithms. The Existing Systems achieved an accuracy of 90% using Random forest, 89% using Logistic regression, 89% using Naive Bayes, and 97.5 using Deep Neural Networks. The Proposed model achieved an accuracy of 100% using Random forest, 97% using Logistic regression,90% using Naive Bayes, and 98.4% using Deep Neural Networks. An increase in accuracy can be achieved using a proposed model that combines screening tool rules with ML algorithms. All existing systems for detection have used ML algorithms and not included screening tool rules which play an important role in the detection process. The Recommendation System is a Monolithic Hybrid model that seamlessly integrates content-based and multi-criteria collaborative filtering. The latter generates cohorts of similar children based on multiple similarity metrics, feeding this information into the content-based filtering component. The output of content-based filtering, in turn, serves as an input to an expert system, which formulates therapy recommendations with prioritization based on the severity of symptoms. This innovative combination enhances the precision and personalization of therapy recommendations for autistic children, marking a significant advancement in the field. This approach surpasses the performance of conventional content-based and collaborative filtering recommender systems, ensuring precise and suitable recommendations with a precision of 80%. This detection and recommendation system helps therapists to accurately detect autism and provide therapies of the highest priority for the wellness of children rather than doing it manually.