Japanese
English
- 有料閲覧
- Abstract 文献概要
- 1ページ目 Look Inside
- 参考文献 Reference
要旨●機械学習(ニューラルネットワーク)を用いて,早期胃癌内視鏡的根治度C-2病変のリンパ節転移予測モデルを作成した.内視鏡的切除および外科的切除後,病理診断が内視鏡的根治度C-2の基準を満たした早期胃癌4,042例の臨床病理学的データを収集し,それらを訓練用コホート(3,506例)と検証用コホート(536例)に分割した.訓練用コホートを用いてリンパ節転移予測モデルを作成し,検証用コホートにおけるその予測精度(AUC)をeCura systemと比較した.モデルの学習には,病変径・主要組織型・組織型混在・深達度・リンパ管侵襲・静脈侵襲・治療法の7因子を用いた.検証用コホートにおけるAUCは,eCura systemの0.77に対し,本モデルは0.83で有意に高い結果が得られた(p=0.006).
We developed a machine learning(ML)model to predict the risk of lymph node metastasis(LNM)in patients with early gastric cancer(EGC)not meeting the current Japanese endoscopic curability criteria. We used the data of 4,042 EGC patients who were histologically confirmed not to meet the abovementioned criteria. Altogether, 3,506 patients were assigned to the training cohort for developing the neural network-based ML model, and 536 patients were included in the validation cohort. The performance of our ML model, as measured by the area under the receiver operating characteristic curve(AUC), was compared with that of the eCura system in the validation cohort. The ML model and eCura system identified patients with LNM with AUC values of 0.83 and 0.77, respectively, in the validation cohort(p=0.006).

Copyright © 2025, Igaku-Shoin Ltd. All rights reserved.