{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"medRxiv"}],"indexed":{"date-parts":[[2026,7,29]],"date-time":"2026-07-29T19:58:19Z","timestamp":1785355099275,"version":"3.55.0"},"posted":{"date-parts":[[2026,7,27]]},"group-title":"Health Economics","reference-count":20,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2026,7,27]],"date-time":"2026-07-27T00:00:00Z","timestamp":1785110400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,7,27]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:p>FeePredict is a three-stage random forest machine learning framework to simultaneously predict whether Medicare reimbursement rates for specific procedures will change, in which direction they will change, and by how much. FeePredict was applied to the four major Medicare fee schedules: the Clinical Laboratory Fee Schedule (CLFS), the Physician Fee Schedule (PFS), the Ambulance Fee Schedule (AFS), and the Durable Medical Equipment, Prosthetics, Orthotics, and Supplies (DMEPOS) fee schedule. Each of these fee schedules contains publicly available data from the Centers for Medicare &amp; Medicaid Services (CMS) for the years 2024, 2025, and 2026, with the number of procedures represented in the data ranging from 3,264 to 2,952,842 observations.<\/jats:p>\n                <jats:p>\n                  FeePredict utilizes lag-1 feature engineering and train-only preprocessing steps to ensure that there is no data leakage into the model. Chronological out-of-time validation was performed on three of the four fee schedules to determine the generalizability of the model over time. FeePredict significantly outperformed the assumption that there would be no changes to Medicare reimbursement rates for procedures (\n                  <jats:italic>p &lt;<\/jats:italic>\n                  0.001), achieving concordance indices between 0.815 and 0.998, and reducing the mean absolute error for predicting changes to reimbursement rates by 29% to 85%.\n                <\/jats:p>\n                <jats:p>Permutation testing of the model with shuffled reimbursement rate labels indicates that there is no evidence of data leakage (AUC values: 0.467-0.515). The model achieved concordance indices of 0.854 and 0.972 for the CLFS and DMEPOS fee schedules, respectively, outside of its training period, but performed less well outside of its training period for the PFS, indicating that it generalizes less well to changes to the Medicare policy regime that existed after its training period. Overall, though, these results indicate that it is possible to accurately predict whether Medicare reimbursement rates for medical procedures will change using only data from the historical versions of those fee schedules.<\/jats:p>","DOI":"10.64898\/2026.07.22.26358736","type":"posted-content","created":{"date-parts":[[2026,7,27]],"date-time":"2026-07-27T17:25:20Z","timestamp":1785173120000},"source":"Crossref","is-referenced-by-count":0,"title":["A Multi-Schedule Machine Learning Pipeline for Medicare Reimbursement Change Prediction and Operational Risk Stratification"],"prefix":"10.64898","author":[{"given":"Ishan H.","family":"Patel","sequence":"first","affiliation":[{"name":"AI4Path Lab, Department of Pathology, The Ohio State University Wexner Medical Center The Ohio State University, Columbus, OH, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0002-0731-9383","authenticated-orcid":false,"given":"Alejandro","family":"Leyva","sequence":"additional","affiliation":[{"name":"AI4Path Lab, Department of Pathology, The Ohio State University Wexner Medical Center The Ohio State University, Columbus, OH, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Muhammad Khalid Khan","family":"Niazi","sequence":"additional","affiliation":[{"name":"AI4Path Lab, Department of Pathology, The Ohio State University Wexner Medical Center The Ohio State University, Columbus, OH, USA"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2026072912000917000_2026.07.22.26358736v1.1","doi-asserted-by":"publisher","DOI":"10.1001\/jama.1988.03410160021004"},{"key":"2026072912000917000_2026.07.22.26358736v1.2","doi-asserted-by":"publisher","DOI":"10.3174\/ajnr.A3674"},{"key":"2026072912000917000_2026.07.22.26358736v1.3","first-page":"1312","article-title":"Supervised learning methods for predicting healthcare costs: systematic literature review and empirical evaluation","volume":"2017","year":"2017","journal-title":"AMIA Annu Symp Proc"},{"key":"2026072912000917000_2026.07.22.26358736v1.4","doi-asserted-by":"crossref","unstructured":"Grinsztajn L , Oyallon E , Varoquaux G . Why do tree-based models still outperform deep learning on typical tabular data? Adv Neural Inf Process Syst. 2022;35.","DOI":"10.52202\/068431-0037"},{"key":"2026072912000917000_2026.07.22.26358736v1.5","doi-asserted-by":"crossref","first-page":"84","DOI":"10.1016\/j.inffus.2021.11.011","article-title":"Tabular data: Deep learning is not all you need","volume":"81","year":"2022","journal-title":"Inf Fusion"},{"key":"2026072912000917000_2026.07.22.26358736v1.6","doi-asserted-by":"publisher","DOI":"10.1287\/opre.1080.0619"},{"key":"2026072912000917000_2026.07.22.26358736v1.7","doi-asserted-by":"publisher","DOI":"10.1111\/1475-6773.12464"},{"key":"2026072912000917000_2026.07.22.26358736v1.8","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0213258"},{"issue":"2","key":"2026072912000917000_2026.07.22.26358736v1.9","doi-asserted-by":"crossref","first-page":"e70093","DOI":"10.1111\/1475-6773.70093","article-title":"A machine learning model to improve risk adjustment accuracy in Medicare","volume":"61","year":"2026","journal-title":"Health Serv Res"},{"issue":"10","key":"2026072912000917000_2026.07.22.26358736v1.10","doi-asserted-by":"crossref","first-page":"e2020291","DOI":"10.1001\/jamanetworkopen.2020.20291","article-title":"Use of data-driven methods to predict long-term patterns of health care spending for Medicare patients","volume":"3","year":"2020","journal-title":"JAMA Netw Open"},{"issue":"1","key":"2026072912000917000_2026.07.22.26358736v1.11","first-page":"131","article-title":"Machine learning approaches for predicting high cost high need patient expenditures in health care","volume":"17","year":"2018","journal-title":"Biomed Eng Online"},{"issue":"5","key":"2026072912000917000_2026.07.22.26358736v1.12","doi-asserted-by":"crossref","first-page":"e22461","DOI":"10.2196\/22461","article-title":"Neural machine translation-based automated current procedural terminology classification system using procedure text","volume":"5","year":"2021","journal-title":"JMIR Form Res"},{"issue":"1","key":"2026072912000917000_2026.07.22.26358736v1.13","first-page":"5","article-title":"Development of a model for the validation of work relative value units for the Medicare Physician Fee Schedule","volume":"5","year":"2015","journal-title":"Rand Health Q"},{"key":"2026072912000917000_2026.07.22.26358736v1.14","first-page":"102752","article-title":"Physician responses to Medicare reimbursement rates","volume":"89","year":"2023","journal-title":"J Health Econ"},{"issue":"10","key":"2026072912000917000_2026.07.22.26358736v1.15","first-page":"qxaf189","article-title":"Why the Medicare physician fee schedule misvalues fee levels and how to fix it","volume":"3","year":"2025","journal-title":"Health Aff Sch"},{"key":"2026072912000917000_2026.07.22.26358736v1.16","first-page":"103568","article-title":"Learning hidden patterns from patient multivariate time series data using convolutional neural networks: a case study of healthcare cost prediction","volume":"111","year":"2020","journal-title":"J Biomed Inform"},{"issue":"Suppl 1","key":"2026072912000917000_2026.07.22.26358736v1.17","first-page":"3189","article-title":"Mental health risk adjustment with clinical categories and machine learning","volume":"53","year":"2018","journal-title":"Health Serv Res"},{"issue":"7","key":"2026072912000917000_2026.07.22.26358736v1.18","doi-asserted-by":"crossref","first-page":"1763","DOI":"10.1007\/s11606-021-07333-z","article-title":"Prediction models for future high-need high-cost healthcare use: a systematic review","volume":"37","year":"2022","journal-title":"J Gen Intern Med"},{"issue":"1","key":"2026072912000917000_2026.07.22.26358736v1.19","doi-asserted-by":"crossref","first-page":"125","DOI":"10.1186\/s12902-020-00609-1","article-title":"Not there yet: using data-driven methods to predict who becomes costly among low-cost patients with type 2 diabetes","volume":"20","year":"2020","journal-title":"BMC Endocr Disord"},{"issue":"7","key":"2026072912000917000_2026.07.22.26358736v1.20","doi-asserted-by":"crossref","first-page":"2022","DOI":"10.1177\/21925682231164935","article-title":"Can a novel natural language processing model and artificial intelligence automatically generate billing codes from spine surgical operative notes?","volume":"14","year":"2024","journal-title":"Global Spine J"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.64898\/2026.07.22.26358736","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,29]],"date-time":"2026-07-29T19:00:42Z","timestamp":1785351642000},"score":1,"resource":{"primary":{"URL":"http:\/\/medrxiv.org\/lookup\/doi\/10.64898\/2026.07.22.26358736"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,27]]},"references-count":20,"URL":"https:\/\/doi.org\/10.64898\/2026.07.22.26358736","relation":{},"subject":[],"published":{"date-parts":[[2026,7,27]]},"subtype":"preprint"}}