{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,30]],"date-time":"2026-07-30T15:19:22Z","timestamp":1785424762545,"version":"3.56.0"},"reference-count":34,"publisher":"MDPI AG","issue":"13","license":[{"start":{"date-parts":[[2026,6,28]],"date-time":"2026-06-28T00:00:00Z","timestamp":1782604800000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"DOI":"10.13039\/501100001809","name":"National Natural Science Foundation of China","doi-asserted-by":"publisher","award":["82472073"],"award-info":[{"award-number":["82472073"]}],"id":[{"id":"10.13039\/501100001809","id-type":"DOI","asserted-by":"publisher"}]},{"award":["82472073"],"award-info":[{"award-number":["82472073"]}],"id":[{"id":"https:\/\/ror.org\/01h0zpd94","id-type":"ROR","asserted-by":"publisher"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Diagnostics"],"abstract":"<jats:p>Objective: The objective of this study was to construct a large language model (LLM) for the Response Evaluation Criteria in Solid Tumors (RECIST) assessment using exclusively longitudinal radiology report text. Methods: This study included 258 patients with solid tumors, encompassing 2065 longitudinal CT\/MRI examination time points. We developed WiNGPT-32B, an open-source and locally deployable LLM, by infusing it with domain-specific medical knowledge and optimizing it via knowledge distillation, using GPT-4 as the teacher model. Central to its architecture is the Chained Task Execution (CTE) framework, which structures RECIST assessment into four modular components: lesion diameter extraction, sum of longest diameter computation, tumor response classification, and report generation. Model performance (accuracy, recall, precision, and F1 score) was benchmarked against GPT-4 and a single radiologist, utilizing the consensus of three independent radiologists as the reference standard. Results: The number of patients with imaging time points was 212 (82.2%) with 4\u201310, 36 (13.9%) with 11\u201320, and 10 (3.9%) with &gt;20 time points. For target lesions, the successful extraction rate of WiNGPT-32B was 0.934 (95% CI: 0.922\u20130.944), which was slightly higher than that of GPT-4 0.920 (0.907\u20130.931; p = 0.083). In five-category RECIST classification (complete response, partial response, stable disease, progressive disease, and not evaluable), WiNGPT-32B achieved an overall accuracy of 0.805 (0.786\u20130.823), significantly higher than GPT-4 (0.699, 0.678\u20130.720; p &lt; 0.001) but lower than the radiologist (0.915, 0.901\u20130.928; p &lt; 0.001). For progressive disease, WiNGPT-32B had an F1 score of 0.841 (0.813\u20130.870), significantly outperforming GPT-4\u2019s 0.755 (0.720\u20130.790), and approaching the radiologist\u2019s 0.922 (0.902\u20130.942). Conclusions: WiNGPT-32B demonstrates the feasibility of a text-only, open-source LLM with the CTE framework for longitudinal RECIST assessment, with promising performance in detecting disease progression.<\/jats:p>","DOI":"10.3390\/diagnostics16132020","type":"journal-article","created":{"date-parts":[[2026,6,29]],"date-time":"2026-06-29T02:16:44Z","timestamp":1782699404000},"page":"2020","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["WiNGPT-32B: An Open-Source, Locally Deployable LLM for RECIST Assessment via Chained Task Execution Using Radiology Report Text"],"prefix":"10.3390","volume":"16","author":[{"given":"Lingyun","family":"Wang","sequence":"first","affiliation":[{"name":"Radiology Department, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Haining Rd. 100, Shanghai 200080, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lu","family":"Zhang","sequence":"additional","affiliation":[{"name":"Radiology Department, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Haining Rd. 100, Shanghai 200080, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Yaping","family":"Zhang","sequence":"additional","affiliation":[{"name":"Radiology Department, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Haining Rd. 100, Shanghai 200080, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Lin","family":"Zhang","sequence":"additional","affiliation":[{"name":"Radiology Department, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Haining Rd. 100, Shanghai 200080, China"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-6669-0097","authenticated-orcid":false,"given":"Xueqian","family":"Xie","sequence":"additional","affiliation":[{"name":"Radiology Department, Shanghai General Hospital, Shanghai Jiao Tong University School of Medicine, Haining Rd. 100, Shanghai 200080, China"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,6,28]]},"reference":[{"key":"ref_1","first-page":"10","article-title":"Cancer statistics, 2025","volume":"75","author":"Siegel","year":"2025","journal-title":"CA Cancer J. Clin."},{"key":"ref_2","doi-asserted-by":"crossref","first-page":"725","DOI":"10.1038\/s41571-024-00932-9","article-title":"Tumour mutational burden: Clinical utility, challenges and emerging improvements","volume":"21","author":"Budczies","year":"2024","journal-title":"Nat. Rev. Clin. Oncol."},{"key":"ref_3","doi-asserted-by":"crossref","unstructured":"Ng, L., Navarro, A., and Law, W.L. (2023). Editorial: Evolving roles of piRNAs in solid tumors. Front. Oncol., 13.","DOI":"10.3389\/fonc.2023.1178634"},{"key":"ref_4","doi-asserted-by":"crossref","first-page":"1003","DOI":"10.1016\/j.ccell.2023.04.012","article-title":"RECISTv1.1 progression in oncology: Shades of gray","volume":"41","author":"Topp","year":"2023","journal-title":"Cancer Cell"},{"key":"ref_5","doi-asserted-by":"crossref","first-page":"228","DOI":"10.1016\/j.ejca.2008.10.026","article-title":"New response evaluation criteria in solid tumours: Revised RECIST guideline (version 1.1)","volume":"45","author":"Eisenhauer","year":"2009","journal-title":"Eur. J. Cancer"},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"e2255113","DOI":"10.1001\/jamanetworkopen.2022.55113","article-title":"Comparison of chest radiograph captions based on natural language processing vs. completed by radiologists","volume":"6","author":"Zhang","year":"2023","journal-title":"JAMA Netw. Open"},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"2258","DOI":"10.1038\/s41467-025-57426-0","article-title":"Towards a holistic framework for multimodal LLM in 3D brain CT radiology report generation","volume":"16","author":"Li","year":"2025","journal-title":"Nat. Commun."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"e231593","DOI":"10.1148\/radiol.231593","article-title":"Quantitative evaluation of large language models to streamline radiology report impressions: A multimodal retrospective analysis","volume":"310","author":"Doshi","year":"2024","journal-title":"Radiology"},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"e241554","DOI":"10.1148\/radiol.241554","article-title":"Accuracy of large language model-based automatic calculation of ovarian-adnexal reporting and data system MRI scores from pelvic MRI reports","volume":"315","author":"Bhayana","year":"2025","journal-title":"Radiology"},{"key":"ref_10","doi-asserted-by":"crossref","unstructured":"Kim, J., Podlasek, A., Shidara, K., Liu, F., Alaa, A., and Bernardo, D. (2025). Limitations of large language models in clinical problem-solving arising from inflexible reasoning. arXiv.","DOI":"10.1038\/s41598-025-22940-0"},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"e240885","DOI":"10.1148\/radiol.240885","article-title":"Constructing a large language model to generate impressions from findings in radiology reports","volume":"312","author":"Zhang","year":"2024","journal-title":"Radiology"},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"Shool, S., Adimi, S., Saboori Amleshi, R., Bitaraf, E., Golpira, R., and Tara, M. (2025). A systematic review of large language model (LLM) evaluations in clinical medicine. BMC Med. Inform. Decis. Mak., 25.","DOI":"10.1186\/s12911-025-02954-4"},{"key":"ref_13","first-page":"e79221","article-title":"DeepSeek in healthcare: Revealing opportunities and steering challenges of a new open-source artificial intelligence frontier","volume":"17","author":"Temsah","year":"2025","journal-title":"Cureus"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"2550","DOI":"10.1038\/s41591-025-03726-3","article-title":"Comparative benchmarking of the DeepSeek large language model on medical tasks and clinical reasoning","volume":"31","author":"Tordjman","year":"2025","journal-title":"Nat. Med."},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"10076","DOI":"10.1109\/TPAMI.2024.3435571","article-title":"Medical image segmentation review: The success of U-Net","volume":"46","author":"Azad","year":"2024","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_16","doi-asserted-by":"crossref","first-page":"127063","DOI":"10.1016\/j.neucom.2023.127063","article-title":"Roformer: Enhanced transformer with rotary position embedding","volume":"568","author":"Su","year":"2024","journal-title":"Neurocomputing"},{"key":"ref_17","unstructured":"Zhang, B., and Sennrich, R. (2019). Root mean square layer normalization. arXiv."},{"key":"ref_18","unstructured":"Shazeer, N. (2020). Glu variants improve transformer. arXiv."},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"7035","DOI":"10.1109\/TPAMI.2020.3001940","article-title":"Structured knowledge distillation for dense prediction","volume":"45","author":"Liu","year":"2023","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref_20","unstructured":"Zhuang, B., Song, C., Lu, H., Qiao, J., Liu, M., Yu, M., Hong, P., Li, R., Song, X., and Xu, X. (2025). WiNGPT-3.0 Technical Report. arXiv."},{"key":"ref_21","first-page":"216","article-title":"Workflow resource patterns: Identification, representation and tool support","volume":"Volume 3520","author":"Russell","year":"2005","journal-title":"Advanced Information Systems Engineering"},{"key":"ref_22","unstructured":"Wu, Y., Wang, Y., Du, T., Jegelka, S., and Wang, Y. (2025). When more is less: Understanding chain-of-thought length in LLMs. arXiv."},{"key":"ref_23","doi-asserted-by":"crossref","first-page":"66","DOI":"10.1186\/s13244-025-01941-7","article-title":"Generative pre-trained transformer 4o (GPT-4o) in solving text-based multiple response questions for European Diploma in Radiology (EDiR): A comparative study with radiologists","volume":"16","author":"Pristoupil","year":"2025","journal-title":"Insights Imaging"},{"key":"ref_24","first-page":"65","article-title":"A simple sequentially rejective multiple test procedure","volume":"6","author":"Holm","year":"1979","journal-title":"Scand. J. Stat."},{"key":"ref_25","doi-asserted-by":"crossref","first-page":"159","DOI":"10.2307\/2529310","article-title":"The measurement of observer agreement for categorical data","volume":"33","author":"Landis","year":"1977","journal-title":"Biometrics"},{"key":"ref_26","doi-asserted-by":"crossref","unstructured":"Mottin, L., Goldman, J.P., J\u00e4ggli, C., Achermann, R., Gobeill, J., Knafou, J., Ehrsam, J., Wicky, A., G\u00e9rard, C.L., and Schwenk, T. (2023). Multilingual RECIST classification of radiology reports using supervised learning. Front. Digit. Health, 5.","DOI":"10.3389\/fdgth.2023.1195017"},{"key":"ref_27","doi-asserted-by":"crossref","first-page":"20552076251394622","DOI":"10.1177\/20552076251394622","article-title":"Clinical reasoning from real-world oncology reports using large language models","volume":"11","author":"Park","year":"2025","journal-title":"Digit. Health"},{"key":"ref_28","doi-asserted-by":"crossref","unstructured":"Leszczy\u0144ska, A., Seweryn, M., Obuchowicz, R., Strzelecki, M., Pi\u00f3rkowski, A., and Potocki, P.M. (2026). Artificial Intelligence for RECIST-Based Radiologic Treatment Response Assessment in Solid Tumors: A Systematic Review of Imaging- and Report-Derived Approaches. Cancers, 18.","DOI":"10.3390\/cancers18050808"},{"key":"ref_29","doi-asserted-by":"crossref","first-page":"1669","DOI":"10.1097\/JS9.0000000000001989","article-title":"Advancements in AI-driven oncology: Assessing ChatGPT\u2019s impact from GPT-3.5 to GPT-4o","volume":"111","author":"Mu","year":"2025","journal-title":"Int. J. Surg."},{"key":"ref_30","doi-asserted-by":"crossref","unstructured":"Shahriar, S., Lund, B.D., Mannuru, N.R., Arshad, M.A., Hayawi, K., Bevara, R.V.K., Mannuru, A., and Batool, L. (2024). Putting GPT-4o to the sword: A comprehensive evaluation of language, vision, speech, and multimodal proficiency. Appl. Sci., 14.","DOI":"10.20944\/preprints202406.1635.v1"},{"key":"ref_31","doi-asserted-by":"crossref","first-page":"833","DOI":"10.1038\/s43588-023-00527-x","article-title":"Human-like intuitive behavior and reasoning biases emerged in large language models but disappeared in ChatGPT","volume":"3","author":"Hagendorff","year":"2023","journal-title":"Nat. Comput. Sci."},{"key":"ref_32","doi-asserted-by":"crossref","first-page":"451","DOI":"10.1038\/s41586-025-08869-4","article-title":"Towards accurate differential diagnosis with large language models","volume":"642","author":"McDuff","year":"2025","journal-title":"Nature"},{"key":"ref_33","doi-asserted-by":"crossref","first-page":"397","DOI":"10.1038\/s41746-025-01780-2","article-title":"Large language model trained on clinical oncology data predicts cancer progression","volume":"8","author":"Zhu","year":"2025","journal-title":"npj Digit. Med."},{"key":"ref_34","doi-asserted-by":"crossref","first-page":"e2036741","DOI":"10.1001\/jamanetworkopen.2020.36741","article-title":"Comparison of Solid Tumor Treatment Response Observed in Clinical Practice With Response Reported in Clinical Trials","volume":"4","author":"Feinberg","year":"2021","journal-title":"JAMA Netw. Open"}],"container-title":["Diagnostics"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2075-4418\/16\/13\/2020\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,7]],"date-time":"2026-07-07T04:18:31Z","timestamp":1783397911000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2075-4418\/16\/13\/2020"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,6,28]]},"references-count":34,"journal-issue":{"issue":"13","published-online":{"date-parts":[[2026,7]]}},"alternative-id":["diagnostics16132020"],"URL":"https:\/\/doi.org\/10.3390\/diagnostics16132020","relation":{},"ISSN":["2075-4418"],"issn-type":[{"value":"2075-4418","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,6,28]]}}}