{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,6]],"date-time":"2025-12-06T09:13:57Z","timestamp":1765012437408,"version":"3.46.0"},"reference-count":36,"publisher":"Wiley","issue":"27-28","license":[{"start":{"date-parts":[[2025,10,29]],"date-time":"2025-10-29T00:00:00Z","timestamp":1761696000000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/onlinelibrary.wiley.com\/termsAndConditions#vor"}],"content-domain":{"domain":["onlinelibrary.wiley.com"],"crossmark-restriction":true},"short-container-title":["Concurrency and Computation"],"published-print":{"date-parts":[[2025,12,25]]},"abstract":"<jats:title>ABSTRACT<\/jats:title>\n                  <jats:p>Feature selection in clinical applications requires domain expertise to identify clinically meaningful predictors, presenting challenges for traditional statistical methods that struggle to incorporate semantic relationships and clinical reasoning. This study evaluates Large Language Model (LLM)\u2010based feature selection methods for drug recommendation tasks, investigating their potential to leverage pre\u2010trained clinical knowledge compared to traditional algorithmic approaches. We systematically evaluated six state\u2010of\u2010the\u2010art LLMs (GPT\u20104, GPT\u20103.5 Turbo, LLaMA variants, and DeepSeek R1) across multiple prompting strategies including rank\u2010based, score\u2010based, Chain\u2010of\u2010Thought reasoning, and a novel Tree\u2010of\u2010Thoughts for Feature Selection (ToT\u2010FS) framework. All methods were evaluated on the MIMIC\u2010III clinical database for drug recommendation, with downstream performance assessed using XGBoost classification. Performance was compared against traditional methods including MRMR, LASSO, and random selection baselines. Advanced LLMs achieved competitive performance with traditional methods, with macro\u2010F1 scores exceeding 0.95 through optimized prompting strategies, closely approaching MRMR (0.97) and LASSO (0.97) performance. Score\u2010based prompting significantly outperformed rank\u2010based approaches, improving GPT\u20104 from 0.83 to 0.95 macro\u2010F1 score. The ToT\u2010FS framework demonstrated consistent high performance across different search depths. Notably, substantial performance disparities emerged between model generations, with older LLaMA 2 variants showing near\u2010random performance (F1\u2009&lt;\u20090.03), while advanced models demonstrated emergent clinical reasoning capabilities. LLM\u2010based feature selection represents a promising paradigm for clinical applications, offering competitive performance with traditional methods while providing enhanced interpretability and domain knowledge integration. The emergence of effective clinical feature selection capabilities at specific model scales suggests advanced LLMs have internalized substantial clinical knowledge, positioning them as valuable tools for clinical decision support systems requiring transparent, interpretable feature selection.<\/jats:p>","DOI":"10.1002\/cpe.70370","type":"journal-article","created":{"date-parts":[[2025,10,29]],"date-time":"2025-10-29T18:36:52Z","timestamp":1761763012000},"update-policy":"https:\/\/doi.org\/10.1002\/crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Leveraging Large Language Models for Feature Selection in Drug Recommendation Systems"],"prefix":"10.1002","volume":"37","author":[{"given":"Sondes","family":"Dardour","sequence":"first","affiliation":[{"name":"University of Gabes, ISGGB  Gabes Tunisia"}],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Nacim","family":"Yanes","sequence":"additional","affiliation":[{"name":"University of Gabes, ISGGB  Gabes Tunisia"},{"name":"University of Manouba, ENSI, RIADI (LR99ES26)  Manouba Tunisia"}],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"311","published-online":{"date-parts":[[2025,10,29]]},"reference":[{"key":"e_1_2_8_2_1","doi-asserted-by":"crossref","unstructured":"S.Imani L.Du andH.Shrivastava \u201cMathprompter: Mathematical Reasoning Using Large Language Models. arXiv preprint arXiv:2303.05398 \u201d2023.","DOI":"10.18653\/v1\/2023.acl-industry.4"},{"key":"e_1_2_8_3_1","unstructured":"R.Taylor M.Kardas G.Cucurull et al. \u201cGalactica: A Large Language Model for Science. arXiv preprint arXiv:2211.09085 \u201d2022."},{"key":"e_1_2_8_4_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-023-06291-2"},{"key":"e_1_2_8_5_1","doi-asserted-by":"crossref","unstructured":"Y.Tong D.Li S.Wang Y.Wang F.Teng andJ.Shang \u201cCan LLMs Learn From Previous Mistakes? Investigating Llms' Errors to Boost for Reasoning. arXiv preprint arXiv:2403.20046 \u201d2024.","DOI":"10.18653\/v1\/2024.acl-long.169"},{"key":"e_1_2_8_6_1","unstructured":"X.Wang J.Wei D.Schuurmans et al. \u201cSelf\u2010Consistency Improves Chain of Thought Reasoning in Language Models. arXiv preprint arXiv:2203.11171 \u201d2022."},{"key":"e_1_2_8_7_1","doi-asserted-by":"publisher","DOI":"10.1109\/TKDE.2024.3352100"},{"key":"e_1_2_8_8_1","doi-asserted-by":"publisher","DOI":"10.1007\/s41019-025-00285-y"},{"key":"e_1_2_8_9_1","doi-asserted-by":"publisher","DOI":"10.18653\/v1\/2023.acl-long.868"},{"key":"e_1_2_8_10_1","unstructured":"W. X.Zhao K.Zhou J.Li et al. \u201cA Survey of Large Language Models. arXiv preprint arXiv:2303.18223 1 2 \u201d2023."},{"key":"e_1_2_8_11_1","doi-asserted-by":"publisher","DOI":"10.1162\/153244303322753616"},{"key":"e_1_2_8_12_1","doi-asserted-by":"publisher","DOI":"10.1142\/S0219720005001004"},{"key":"e_1_2_8_13_1","doi-asserted-by":"publisher","DOI":"10.1111\/j.2517-6161.1996.tb02080.x"},{"key":"e_1_2_8_14_1","doi-asserted-by":"publisher","DOI":"10.1093\/bioinformatics\/btm344"},{"key":"e_1_2_8_15_1","doi-asserted-by":"publisher","DOI":"10.1016\/j.ijmedinf.2006.11.006"},{"key":"e_1_2_8_16_1","doi-asserted-by":"publisher","DOI":"10.1038\/nrg3208"},{"key":"e_1_2_8_17_1","doi-asserted-by":"publisher","DOI":"10.1038\/s41591-023-02448-8"},{"key":"e_1_2_8_18_1","unstructured":"D. P.Jeong Z. C.Lipton andP.Ravikumar \u201cLlm\u2010Select: Feature Selection With Large Language Models. arXiv preprint arXiv:2407.02694 \u201d2024."},{"key":"e_1_2_8_19_1","doi-asserted-by":"publisher","DOI":"10.1145\/3136625"},{"key":"e_1_2_8_20_1","unstructured":"D.ShareefandG. A.Yosefi \u201cA Literature Review of Feature Selection Methods. EasyChair Preprint \u201d2021."},{"key":"e_1_2_8_21_1","doi-asserted-by":"publisher","DOI":"10.3115\/1075527.1075574"},{"key":"e_1_2_8_22_1","first-page":"1393","article-title":"Feature selection via dependence maximization","volume":"13","author":"Song L.","year":"2012","journal-title":"Journal of Machine Learning Research"},{"key":"e_1_2_8_23_1","doi-asserted-by":"publisher","DOI":"10.1023\/A:1012487302797"},{"key":"e_1_2_8_24_1","unstructured":"K.Choi C.Cundy S.Srivastava andS.Ermon \u201cLmpriors: Pre\u2010Trained Language Models as Task\u2010Specific Priors. arXiv preprint arXiv:2210.12530 \u201d2022."},{"key":"e_1_2_8_25_1","doi-asserted-by":"publisher","DOI":"10.1145\/3715073.3715077"},{"key":"e_1_2_8_26_1","unstructured":"J.LiandX.Xiu \u201cLLM4FS: Leveraging Large Language Models for Feature Selection and How to Improve It. arXiv preprint arXiv:2503.24157 \u201d2025."},{"key":"e_1_2_8_27_1","doi-asserted-by":"publisher","DOI":"10.1145\/3560815"},{"key":"e_1_2_8_28_1","unstructured":"B.Mann N.Ryder M.Subbiah et al. \u201cLanguage Models Are Few\u2010Shot Learners. arXiv preprint arXiv:2005.14165 1: 3 \u201d2020."},{"key":"e_1_2_8_29_1","unstructured":"A.Srivastava A.Rastogi A.Rao et al. \u201cBeyond the Imitation Game: Quantifying and Extrapolating the Capabilities of Language Models. arXiv preprint arXiv:2206.04615 \u201d2022."},{"key":"e_1_2_8_30_1","doi-asserted-by":"crossref","unstructured":"S.Lin J.Hilton andO.Evans \u201cTruthfulqa: Measuring How Models Mimic Human Falsehoods. arXiv preprint arXiv:2109.07958 \u201d2021.","DOI":"10.18653\/v1\/2022.acl-long.229"},{"key":"e_1_2_8_31_1","unstructured":"R.PatelandE.Pavlick \u201cMapping Language Models to Grounded Conceptual Spaces \u201d2022."},{"key":"e_1_2_8_32_1","doi-asserted-by":"publisher","DOI":"10.1162\/tacl_a_00324"},{"key":"e_1_2_8_33_1","doi-asserted-by":"crossref","unstructured":"T.SchickandH.Sch\u00fctze \u201cIt's Not Just Size That Matters: Small Language Models Are Also Few\u2010Shot Learners. arXiv preprint arXiv:2009.07118 \u201d2020.","DOI":"10.18653\/v1\/2021.naacl-main.185"},{"key":"e_1_2_8_34_1","first-page":"12697","volume-title":"PMLR","author":"Zhao Z.","year":"2021"},{"key":"e_1_2_8_35_1","first-page":"24824","article-title":"Chain\u2010Of\u2010Thought Prompting Elicits Reasoning in Large Language Models","volume":"35","author":"Wei J.","year":"2022","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_2_8_36_1","first-page":"11809","article-title":"Tree of Thoughts: Deliberate Problem Solving With Large Language Models","volume":"36","author":"Yao S.","year":"2023","journal-title":"Advances in Neural Information Processing Systems"},{"key":"e_1_2_8_37_1","unstructured":"J.Long \u201cLarge Language Model Guided Tree\u2010Of\u2010Thought. arXiv preprint arXiv:2305.08291 \u201d2023."}],"container-title":["Concurrency and Computation: Practice and Experience"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/pdf\/10.1002\/cpe.70370","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,6]],"date-time":"2025-12-06T09:10:10Z","timestamp":1765012210000},"score":1,"resource":{"primary":{"URL":"https:\/\/onlinelibrary.wiley.com\/doi\/10.1002\/cpe.70370"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,10,29]]},"references-count":36,"journal-issue":{"issue":"27-28","published-print":{"date-parts":[[2025,12,25]]}},"alternative-id":["10.1002\/cpe.70370"],"URL":"https:\/\/doi.org\/10.1002\/cpe.70370","archive":["Portico"],"relation":{},"ISSN":["1532-0626","1532-0634"],"issn-type":[{"type":"print","value":"1532-0626"},{"type":"electronic","value":"1532-0634"}],"subject":[],"published":{"date-parts":[[2025,10,29]]},"assertion":[{"value":"2025-06-05","order":0,"name":"received","label":"Received","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-10-10","order":2,"name":"accepted","label":"Accepted","group":{"name":"publication_history","label":"Publication History"}},{"value":"2025-10-29","order":3,"name":"published","label":"Published","group":{"name":"publication_history","label":"Publication History"}}],"article-number":"e70370"}}