Result: Machine learning-based disease risk stratification and prediction of metabolic dysfunction-associated fatty liver disease using vibration-controlled transient elastography: Result from NHANES 2021-2023.

Title:
Machine learning-based disease risk stratification and prediction of metabolic dysfunction-associated fatty liver disease using vibration-controlled transient elastography: Result from NHANES 2021-2023.
Authors:
Huang L; Department of Ultrasound, Chengdu Integrated Traditional Chinese Medicine and Western Medicine Hospital, Sichuan Province, No. 18 Wanxiang North Road, High Tech Zone, Chengdu, China., Luo Y; Department of Ultrasound, Chengdu Integrated Traditional Chinese Medicine and Western Medicine Hospital, Sichuan Province, No. 18 Wanxiang North Road, High Tech Zone, Chengdu, China., Zhang L; Department of Ultrasound, Chengdu Integrated Traditional Chinese Medicine and Western Medicine Hospital, Sichuan Province, No. 18 Wanxiang North Road, High Tech Zone, Chengdu, China., Wu M; Department of Ultrasound, Chengdu Integrated Traditional Chinese Medicine and Western Medicine Hospital, Sichuan Province, No. 18 Wanxiang North Road, High Tech Zone, Chengdu, China., Hu L; Department of Ultrasound, Chengdu Integrated Traditional Chinese Medicine and Western Medicine Hospital, Sichuan Province, No. 18 Wanxiang North Road, High Tech Zone, Chengdu, China. 277373164@qq.com.
Source:
BMC gastroenterology [BMC Gastroenterol] 2025 Apr 14; Vol. 25 (1), pp. 255. Date of Electronic Publication: 2025 Apr 14.
Publication Type:
Journal Article
Language:
English
Journal Info:
Publisher: BioMed Central Country of Publication: England NLM ID: 100968547 Publication Model: Electronic Cited Medium: Internet ISSN: 1471-230X (Electronic) Linking ISSN: 1471230X NLM ISO Abbreviation: BMC Gastroenterol Subsets: MEDLINE
Imprint Name(s):
Original Publication: London : BioMed Central, [2001-
MeSH Terms:
Elasticity Imaging Techniques*/methods , Machine Learning* , Non-alcoholic Fatty Liver Disease*/diagnostic imaging , Fatty Liver*/diagnostic imaging, Humans ; Male ; Female ; Middle Aged ; Adult ; Nutrition Surveys ; Risk Assessment/methods ; Vibration ; Risk Factors ; ROC Curve
References:
Ann Nutr Metab. 2022;78(5):247-254. (PMID: 35933983)
Int J Mol Sci. 2022 Oct 18;23(20):. (PMID: 36293337)
Nutrients. 2023 Nov 22;15(23):. (PMID: 38068729)
Hepatology. 2019 Oct;70(4):1457-1469. (PMID: 30924946)
JAMA. 2024 Apr 16;331(15):1287-1297. (PMID: 38512249)
Nutrients. 2023 May 16;15(10):. (PMID: 37242221)
Turk J Gastroenterol. 2021 May;32(5):466-472. (PMID: 34231477)
Curr Issues Mol Biol. 2023 Nov 15;45(11):9132-9148. (PMID: 37998750)
Diabetes Metab Syndr Obes. 2023 Apr 27;16:1221-1234. (PMID: 37139347)
Gastroenterology. 2019 May;156(6):1717-1730. (PMID: 30689971)
Sci Rep. 2023 Jun 6;13(1):9158. (PMID: 37280241)
J Clin Exp Hepatol. 2025 Jan-Feb;15(1):102420. (PMID: 39564428)
Biosci Trends. 2023 Mar 11;17(1):21-37. (PMID: 36682800)
Digit Health. 2024 Aug 7;10:20552076241272535. (PMID: 39119551)
Ther Adv Endocrinol Metab. 2023 Jan 28;14:20420188221145549. (PMID: 36726391)
Nat Rev Gastroenterol Hepatol. 2021 Apr;18(4):223-238. (PMID: 33349658)
Clin Gastroenterol Hepatol. 2009 Oct;7(10):1104-12. (PMID: 19523535)
World J Gastroenterol. 2019 Mar 21;25(11):1307-1326. (PMID: 30918425)
Lipids Health Dis. 2021 Oct 10;20(1):134. (PMID: 34629059)
Front Med (Lausanne). 2024 Nov 13;11:1499013. (PMID: 39606621)
J Gastroenterol Hepatol. 2021 Mar;36(3):543-550. (PMID: 33709607)
Diabetes Care. 2021 Feb;44(2):399-406. (PMID: 33355256)
Front Endocrinol (Lausanne). 2022 May 06;13:848937. (PMID: 35620390)
Lipids Health Dis. 2022 Dec 8;21(1):133. (PMID: 36482400)
Front Endocrinol (Lausanne). 2023 Feb 07;14:1094101. (PMID: 36824362)
J Hepatol. 2020 Jul;73(1):202-209. (PMID: 32278004)
Sci Rep. 2022 Dec 19;12(1):21925. (PMID: 36535977)
Dig Liver Dis. 2024 Apr;56(4):565-570. (PMID: 38104027)
Liver Int. 2025 Mar;45(3):e70042. (PMID: 39999021)
Front Endocrinol (Lausanne). 2022 Nov 04;13:977625. (PMID: 36407325)
J Clin Transl Hepatol. 2022 Apr 28;10(2):197-206. (PMID: 35528980)
Nutrients. 2022 Oct 28;14(21):. (PMID: 36364818)
Front Endocrinol (Lausanne). 2022 Sep 02;13:951689. (PMID: 36120429)
Hepatol Commun. 2022 Jul;6(7):1537-1548. (PMID: 35365931)
J Hepatol. 2021 Oct;75(4):770-785. (PMID: 33991635)
Clin Gastroenterol Hepatol. 2019 Jan;17(1):156-163.e2. (PMID: 29705261)
Metabolism. 2022 Nov;136:155294. (PMID: 35995280)
Diabetes Metab Syndr Obes. 2023 Jul 31;16:2255-2268. (PMID: 37545743)
Front Endocrinol (Lausanne). 2023 Jan 19;14:1083032. (PMID: 36742412)
J Hepatol. 2023 Sep;79(3):842-852. (PMID: 37169151)
Sci Rep. 2024 May 15;14(1):11170. (PMID: 38750109)
BMC Endocr Disord. 2023 Apr 20;23(1):85. (PMID: 37081417)
Sci Rep. 2022 Feb 21;12(1):2914. (PMID: 35190650)
Gut. 2016 Aug;65(8):1359-68. (PMID: 25873639)
World J Hepatol. 2024 Jul 27;16(7):990-994. (PMID: 39086530)
Diabetes Metab. 2024 Jan;50(1):101506. (PMID: 38141808)
Contributed Indexing:
Keywords: Liver fibrosis; Metabolic dysfunction-associated fatty liver disease; Predictive modeling; Risk stratification; Vibration-controlled transient elastography
Entry Date(s):
Date Created: 20250414 Date Completed: 20250414 Latest Revision: 20250417
Update Code:
20250417
PubMed Central ID:
PMC11998142
DOI:
10.1186/s12876-025-03850-x
PMID:
40229697
Database:
MEDLINE

Further Information

Background: Metabolic dysfunction-associated fatty liver disease (MAFLD) is a common chronic liver disease and represents a significant public health issue. Nevertheless, current risk stratification methods remain inadequate. The study aimed to use machine learning in the identification of significant features and the development of a predictive model to determine its usefulness in discrimination of MAFLD's risk stratification (low, moderate, and high) in adults.
Methods: The data of the 2021-2023 NHANES database were analyzed. Vibration-controlled transient elastography measurements, including controlled attenuation parameter for the evaluation of steatosis and liver stiffness for the evaluation of fibrosis, were used for risk stratification. The participants were grouped into low-risk, moderate-risk, and high-risk groups based on specific criteria. Feature selection was conducted through Least Absolute Shrinkage and Selection Operator (LASSO) regression and random forest classification.
Results: A total of 4,227 participants were included in the study. There were 16 significant predictors identified by LASSO regression, among which the top 10 predictors were demographic (age, gender, race, hypertension history), clinical (body mass index, waist circumference, hemoglobin, glycohemoglobin, lymphocyte count), and education level. The area under the receiver operating characteristic curve (AUC) of the random forest model in the validation set was 0.80, and the individual AUC was 0.83, 0.66 and 0.79 for the low-, moderate-, and high-risk groups, respectively.
Conclusion: Our machine learning model has excellent performance in stratification of risk for MAFLD with readily available clinical and demographic parameters. This model could be employed as a valuable screening tool to refer high-risk patients for further hepatological evaluation.
(© 2025. The Author(s).)

Declarations. Ethics approval and consent to participate: The NHANES program received ethical approval from the NCHS Research Ethics Review Board, and all participants provided written informed consent. All participants provided written informed consent. Competing interests: The authors declare no competing interests.