Full
Outline
To develop and validate a risk prediction model for liver cirrhosis applicable to the general population, to reveal the association between common biomarkers and the risk of liver cirrhosis, and to provide a scientific basis for the assessment of liver cirrhosis risk in clinical and public health practice.
This study utilized data on 28 biomarkers and 6 covariates from the UK Biobank. Participants were enrolled between 2006 and 2010 and followed up until October 31, 2022. The primary outcome was new cases of liver cirrhosis. All participants meeting the inclusion and exclusion criteria were randomly divided into a training set and a test set at a ratio of 7:3. Based on the training set, three machine learning algorithms, namely XGBoost, Lasso-Cox regression, and random survival forest, were used to screen for key predictive variables influencing the occurrence of liver cirrhosis. Key feature variables were then used to construct a liver cirrhosis risk prediction model based on the Cox proportional hazards model. The performance of the prediction model was evaluated by comparing it with traditional scores, FIB-4 score and APRI score, and validated in the validation set.
A total of 237 965 eligible subjects were screened for this study, including 1 064 patients with liver cirrhosis and 236 801 non-liver cirrhosis patients. Five key variables influencing the occurrence of liver cirrhosis were identified by combining three machine learning methods and clinical application scenarios, namely γ-glutamyl transferase, aspartate aminotransferase, cystatin C, body mass index, and albumin. The liver cirrhosis risk prediction model constructed based on these five characteristic variables significantly outperformed the traditional FIB-4 score model and APRI score model in terms of predictive performance. The AUC values of the liver cirrhosis risk prediction model, FIB-4 score model, and APRI score model in the training set were 0.827（0.806-0.848）, 0.665（0.637-0.693）, and 0.740（0.713-0.766）, respectively; in the validation set, they were 0.811（0.776-0.845）, 0.664（0.621-0.708）, and 0.711（0.668-0.755）, respectively. The AUC values of the prediction model established in this study were superior to those of the two traditional scoring prediction models, and the differences were statistically significant（P<0.05）. Compared with the traditional FIB-4 score model and APRI score model, the liver cirrhosis risk prediction model constructed in this study demonstrated significant advantages in the two key indicators of net reclassification index（NRI）and integrated discrimination improvement index（IDI）（P<0.001）, and the results were highly consistent in both the training set and the test set. The results of the liver cirrhosis risk prediction model showed that cystatin C had high specificity and predictive value in the prediction of liver fibrosis diseases [HR=2.51, 95% CI: 2.22-2.83].
γ-glutamyl transferase, aspartate aminotransferase, cystatin C, body mass index and albumin are important factors influencing the onset of liver cirrhosis. The prediction model for the risk of liver cirrhosis constructed by using the feature variables obtained through multiple screening by machine learning has strong predictive ability and provides an effective tool for the early screening of the risk of liver cirrhosis.
51
24
| 科 Family | 属数 Number of genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) | 属 Genus | 种数 Number of species | 占总种数比例 Percentage of total species (%) |
|---|---|---|---|---|---|---|
| 鹅膏菌科Amanitaceae | 2 | 11 | 5.26 | 鹅膏菌属 Amanita | 10 | 4.78 |
| 小菇科 Mycenaceae | 2 | 12 | 5.74 | 丝盖伞属 Inocybe | 5 | 2.39 |
| 多孔菌科 Polyporaceae | 8 | 14 | 6.70 | 蜡蘑属 Laccaria | 5 | 2.39 |
| 红菇科 Russulaceae | 3 | 23 | 11.00 | 小皮伞属 Marasmius | 6 | 2.87 |
| 小菇属 Mycena | 11 | 5.26 | ||||
| 光柄菇属 Pluteus | 5 | 2.39 | ||||
| 红菇属 Russula | 17 | 8.13 | ||||
| 栓菌属 Trametes | 5 | 2.39 |
关闭全屏
No. 86 Xueyuan South Road, Haidian District, Beijing
010-62662699(Operation and Technical Support)
qkjq@cast.org.cn
Copyright © 2025 China Association for Science and Technology. All rights reserved. For all open access content, the relevant licensing terms apply.
Sponsored by the Office of the Leading Group for Cybersecurity and Informatization of CAST, and supported by Science and Technology Review Publishing House
