{"doi":"10.1002/jsp2.1178","title":"Optimal machine learning methods for radiomic prediction models: Clinical application for preoperative <scp>T<sub>2</sub></scp>*‐weighted images of cervical spondylotic myelopathy","abstract":"<jats:title>Abstract</jats:title><jats:sec><jats:title>Introduction</jats:title><jats:p>Predicting the postoperative neurological function of cervical spondylotic myelopathy (CSM) patients is generally based on conventional magnetic resonance imaging (MRI) patterns, but this approach is not completely satisfactory. This study utilized radiomics, which produced advanced objective and quantitative indicators, and machine learning to develop, validate, test, and compare models for predicting the postoperative prognosis of CSM.</jats:p></jats:sec><jats:sec><jats:title>Materials and methods</jats:title><jats:p>In total, 151 CSM patients undergoing surgical treatment and preoperative MRI was retrospectively collected and divided into good/poor outcome groups based on postoperative modified Japanese Orthopedic Association (mJOA) scores. The datasets obtained from several scanners (an independent  scanner) for the training (testing) cohort were used for cross‐validation (CV). Radiological models based on the intramedullary hyperintensity and compression ratio were constructed with 14 binary classifiers. Radiomic models based on 237 robust radiomic features were constructed with the same 14 binary classifiers in combination with 7 feature reduction methods, resulting in 98 models. The main outcome measures were the area under the receiver operating characteristic curve (AUROC) and accuracy.</jats:p></jats:sec><jats:sec><jats:title>Results</jats:title><jats:p>Forty‐one (11) radiomic models were superior to random guessing during CV (testing), with significant increased AUROC and/or accuracy (<jats:italic>P</jats:italic><jats:sub>AUROC</jats:sub> &lt; .05 and/or <jats:italic>P</jats:italic><jats:sub>accuracy</jats:sub> &lt; .05). One radiological model performed better than random guessing during CV (<jats:italic>P</jats:italic><jats:sub>accuracy</jats:sub> &lt; .05). In the testing cohort, the linear SVM preprocessor + SVM, the best radiomic model (AUROC: 0.74 ± 0.08, accuracy: 0.73 ± 0.07), overperformed the best radiological model (<jats:italic>P</jats:italic><jats:sub>AUROC</jats:sub> = .048).</jats:p></jats:sec><jats:sec><jats:title>Conclusion</jats:title><jats:p>Radiomic features can predict postoperative spinal cord function in CSM patients. The linear SVM preprocessor + SVM has great application potential in building radiomic models.</jats:p></jats:sec>","journal":"JOR SPINE","year":2021,"id":680615,"datarank":0.42498200160843247,"base_score":2.833213344056216,"endowment":2.833213344056216,"self_citation_contribution":0.42498200160843247,"citation_network_contribution":0.0,"self_endowment_contribution":0.42498200160843247,"citer_contribution":0.0,"corpus_percentile":null,"corpus_rank":null,"citation_count":16,"citer_count":0,"citers_with_citation_signal":0,"citers_with_endowment":0,"datacite_reuse_total":0,"is_dataset":false,"is_dataset_confidence":null,"is_data_producer":false,"deposit_databanks":null,"is_oa":false,"file_count":0,"downloads":0,"has_version_chain":false,"published_date":null,"fair_score":null,"fair_percentile":null,"algorithm_id":"datarank_citation_only_1hop_v6","ranking_scope":"data_only","authors":[{"id":1778263,"name":"Han‐Qiang Ou‐Yang","orcid":null,"position":1,"is_corresponding":false},{"id":1060257,"name":"Liang Jiang","orcid":"0000-0002-7221-2956","position":2,"is_corresponding":false},{"id":1778266,"name":"Chun‐Jie Wang","orcid":null,"position":3,"is_corresponding":false},{"id":1778268,"name":"Jian‐Fang Liu","orcid":null,"position":4,"is_corresponding":false},{"id":234071,"name":"Dan Jin","orcid":null,"position":5,"is_corresponding":false},{"id":78336,"name":"Ming Ni","orcid":"0000-0001-9465-2787","position":6,"is_corresponding":false},{"id":1778271,"name":"Xiao‐Guang Liu","orcid":null,"position":7,"is_corresponding":false},{"id":1778273,"name":"Ning Lang","orcid":null,"position":8,"is_corresponding":false},{"id":1778275,"name":"Hui‐Shu Yuan","orcid":null,"position":9,"is_corresponding":false},{"id":1778261,"name":"Meng‐Ze Zhang","orcid":"0000-0002-2164-1558","position":0,"is_corresponding":false}],"reference_count":0,"raw_metadata":{"has_enrichment":true,"resolved":true,"title":"Optimal machine learning methods for radiomic prediction models: Clinical application for preoperative <scp>T<sub>2</sub></scp>*‐weighted images of cervical spondylotic myelopathy","abstract":"<jats:title>Abstract</jats:title><jats:sec><jats:title>Introduction</jats:title><jats:p>Predicting the postoperative neurological function of cervical spondylotic myelopathy (CSM) patients is generally based on conventional magnetic resonance imaging (MRI) patterns, but this approach is not completely satisfactory. This study utilized radiomics, which produced advanced objective and quantitative indicators, and machine learning to develop, validate, test, and compare models for predicting the postoperative prognosis of CSM.</jats:p></jats:sec><jats:sec><jats:title>Materials and methods</jats:title><jats:p>In total, 151 CSM patients undergoing surgical treatment and preoperative MRI was retrospectively collected and divided into good/poor outcome groups based on postoperative modified Japanese Orthopedic Association (mJOA) scores. The datasets obtained from several scanners (an independent  scanner) for the training (testing) cohort were used for cross‐validation (CV). Radiological models based on the intramedullary hyperintensity and compression ratio were constructed with 14 binary classifiers. Radiomic models based on 237 robust radiomic features were constructed with the same 14 binary classifiers in combination with 7 feature reduction methods, resulting in 98 models. The main outcome measures were the area under the receiver operating characteristic curve (AUROC) and accuracy.</jats:p></jats:sec><jats:sec><jats:title>Results</jats:title><jats:p>Forty‐one (11) radiomic models were superior to random guessing during CV (testing), with significant increased AUROC and/or accuracy (<jats:italic>P</jats:italic><jats:sub>AUROC</jats:sub> &lt; .05 and/or <jats:italic>P</jats:italic><jats:sub>accuracy</jats:sub> &lt; .05). One radiological model performed better than random guessing during CV (<jats:italic>P</jats:italic><jats:sub>accuracy</jats:sub> &lt; .05). In the testing cohort, the linear SVM preprocessor + SVM, the best radiomic model (AUROC: 0.74 ± 0.08, accuracy: 0.73 ± 0.07), overperformed the best radiological model (<jats:italic>P</jats:italic><jats:sub>AUROC</jats:sub> = .048).</jats:p></jats:sec><jats:sec><jats:title>Conclusion</jats:title><jats:p>Radiomic features can predict postoperative spinal cord function in CSM patients. The linear SVM preprocessor + SVM has great application potential in building radiomic models.</jats:p></jats:sec>","is_dataset_classified":null,"base_score":2.833213344056216,"endowment":2.833213344056216,"datacite_reuse_total":0,"file_count":0,"downloads":0,"views":0,"has_version_chain":false,"is_dataset":false,"is_oa":false,"pmid":"35005444","pmcid":"PMC8717093","openalex_id":"https://openalex.org/W3212173160","authors":[],"funders":[{"funder_name":"Natural Science Foundation of Beijing Municipality","grant_id":"BYSY2018003","title":null},{"funder_name":"Natural Science Foundation of Beijing Municipality","grant_id":"BYSYZHKC2020116","title":null},{"funder_name":"Natural Science Foundation of Beijing Municipality","grant_id":"7204327","title":null},{"funder_name":"Natural Science Foundation of Beijing Municipality","grant_id":"Z190020","title":null},{"funder_name":"National Natural Science Foundation of China","grant_id":"82102638","title":null},{"funder_name":"Natural Science Foundation of Beijing Municipality","grant_id":"","title":null}],"total_grants":6,"fwci":1.4168,"citation_percentile":0.8194555,"influential_citations":0,"citation_trend":[{"year":2021,"count":2},{"year":2023,"count":2},{"year":2024,"count":3},{"year":2025,"count":6},{"year":2026,"count":3}],"oa_status":"gold","license":"cc-by-nc-nd","oa_locations":[{"url":"https://doi.org/10.1002/jsp2.1178","host_type":"journal"},{"url":"https://doi.org/10.1002/jsp2.1178","host_type":"GOLD"},{"url":"https://doi.org/10.1002/jsp2.1178","host_type":"publisher"},{"url":"https://onlinelibrary.wiley.com/doi/pdf/10.1002/jsp2.1178","host_type":"publisher"},{"url":"https://onlinelibrary.wiley.com/doi/full-xml/10.1002/jsp2.1178","host_type":"publisher"},{"url":"https://pubmed.ncbi.nlm.nih.gov/35005444","host_type":"repository"},{"url":"https://www.ncbi.nlm.nih.gov/pmc/articles/8717093","host_type":"repository"},{"url":"https://doaj.org/article/30c258de9fa741b2920f8fb195b84752","host_type":"repository"},{"url":"https://europepmc.org/articles/PMC8717093","host_type":"Europe_PMC"},{"url":"https://europepmc.org/articles/PMC8717093?pdf=render","host_type":"Europe_PMC"}],"fields_of_study":["Cervical and Thoracic Myelopathy","Radiomics and Machine Learning in Medical Imaging","Medical Imaging and Analysis","Medicine","Engineering"],"mesh_terms":[],"keywords":["Receiver operating characteristic","Support vector machine","Medicine","Magnetic resonance imaging","Artificial intelligence","Pattern recognition (psychology)","Machine learning","Radiology","Computer science","Cervical Spondylotic Myelopathy","Radiomics"],"sdg_mappings":[{"sdg_number":0,"sdg_label":"No poverty"}],"linked_datasets":[],"clinical_trials":[],"software_tools":[],"database_accessions":[{"name":"doi"}],"source":"live","citation_network_status":"fetched"},"created_at":"2026-08-17T15:36:22.549399Z","pmid":null,"pmcid":null,"fwci":null,"citation_percentile":null,"influential_citations":0,"oa_status":null,"license":null,"views":0,"total_file_size_bytes":0,"version_count":0,"fair_f":null,"fair_a":null,"fair_i":null,"fair_r":null,"fair_zscore":null,"fair_rationale":null,"fair_model":null,"fair_agent_version":null,"fair_fulltext_source":null,"fair_has_llm":null,"fair_computed_at":null,"clinical_trials":[],"software_tools":[],"db_accessions":[],"linked_datasets":[],"topics":[]}