{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/a-comparison-of-machine-learning-algorithms","title":"A Comparison of Machine Learning Algorithms for the Surveillance of Autism Spectrum Disorder","arxiv_id":"1804.06223","date":"2018-04-17","proceeding":null,"authors":["Scott H Lee","Matthew J Maenner","Charles M Heilig"],"abstract":"The Centers for Disease Control and Prevention (CDC) coordinates a\nlabor-intensive process to measure the prevalence of autism spectrum disorder\n(ASD) among children in the United States. Random forests methods have shown\npromise in speeding up this process, but they lag behind human classification\naccuracy by about 5%. We explore whether more recently available document\nclassification algorithms can close this gap. We applied 8 supervised learning\nalgorithms to predict whether children meet the case definition for ASD based\nsolely on the words in their evaluations. We compared the algorithms'\nperformance across 10 random train-test splits of the data, using\nclassification accuracy, F1 score, and number of positive calls to evaluate\ntheir potential use for surveillance. Across the 10 train-test cycles, the\nrandom forest and support vector machine with Naive Bayes features (NB-SVM)\neach achieved slightly more than 87% mean accuracy. The NB-SVM produced\nsignificantly more false negatives than false positives (P = 0.027), but the\nrandom forest did not, making its prevalence estimates very close to the true\nprevalence in the data. The best-performing neural network performed similarly\nto the random forest on both measures. The random forest performed as well as\nmore recently available models like the NB-SVM and the neural network, and it\nalso produced good prevalence estimates. NB-SVM may not be a good candidate for\nuse in a fully-automated surveillance workflow due to increased false\nnegatives. More sophisticated algorithms, like hierarchical convolutional\nneural networks, may not be feasible to train due to characteristics of the\ndata. Current algorithms might perform better if the data are abstracted and\nprocessed differently and if they take into account information about the\nchildren in addition to their evaluations.","url_abs":"http://arxiv.org/abs/1804.06223v3","url_pdf":"http://arxiv.org/pdf/1804.06223v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"a-comparison-of-machine-learning-algorithms","repo_url":"https://github.com/scotthlee/autism_classification","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"machine-learning","task_name":"BIG-bench Machine Learning"},{"task_slug":"document-classification","task_name":"Document Classification"},{"task_slug":"classification","task_name":"General Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}