Diagnosis and Characterization of Non-Alcoholic Fatty Liver Disease Based on Artificial Intelligence. (NASHAI)

A key element in the diagnosis of non-alcoholic fatty liver disease (NAFLD) is the differentiation of non-alcoholic steatohepatitis (NASH) from non-alcoholic fatty liver (NAFL) and the staging of the liver fibrosis, given that patients with NASH and advanced fibrosis are those at greatest risk of developing hepatic complications and cardiovascular disease. There are still no available non-invasive methods that allow for correct diagnosis and staging of NAFLD. The implementation of Artificial Intelligence (AI) techniques based on artificial neural networks and deep learning systems (Deep Learning System) as a tool for medical diagnoses represents a bona fide technological revolution that introduces an innovative approach to improving health processes.

Study Overview

• Status: Unknown
• Conditions: Non-alcoholic Fatty Liver Disease (NAFLD)
• Intervention / Treatment: Other: This is an observational study.

Detailed Description

The objectives of this observational study are the following:

1. To design a predictive model of significant liver disease due to NAFLD, based on clustering or clustering algorithms (AI)
2. To apply and validate this model to classify patients according to the severity of the disease in such a manner as to provide more effective management of these patients from Primary Care to Hospital Care through process and resource optimization
3. To develop a Deep Learning System based on convolutional neuronal networks for automatic recognition of images in a cohort of subjects with digitized liver biopsies, and to undertake pairwise analysis that allows for correct and exact classification of biopsies from subjects with NASH.

Design:

An observational study of the determination and validation of diagnostic predictive models of NAFLD.

The study has four phases:

Phases I and II refer to both unsupervised and supervised artificial intelligence learning to identify clusters and build diagnostic algorithms. They will be carried out on data generated from the ETHON cohort (see below).

Phase III will consist on applying deep learning system technology as a support strategy to stratify liver biopsies in NALFD patients according to their grade of necro-inflammation and stage of fibrosis. Liver biopsies collected in the Spanish registry of NAFLD up to the beginning of the study will be used.

Finally, a phase IV of validation will be performed with data from patients that are going to be registered in the Spanish registry of NAFLD.

Population:

1. - Study cohort (Phases I-III):

   A. Subjects from the general population identified in the ETHON (Epidemiological Study of Hepatic Infections) cohort* that has already been created (12,246 subjects between 19-74 years of age) and B. Subjects belonging to the Spanish registry of NAFLD (HEPAmet) (1,800 subjects already collected at the beginning of the study)

   *The ETHON cohort was recruited between 2015 and 2017 to study the hepatitis C prevalence in the Spanish general population aged 19-74 years old. Lavin AC, Llerena S, Gomez M, Escudero MD, Rodriguez L, Estebanez LA, Gamez B, Puchades L, Cabezas J, Serra MA, Calleja JL, Crespo J. Prevalence of hepatitis C in the spanish population. The PREVHEP study (ETHON cohort). J Hepatol. 2017;66:S272.

2. - Validation cohort (Phase IV):

Patients diagnosed with NAFLD by hepatic biopsy recruited in the Spanish and European registers from the beginning of the study.

-Inclusion and exclusion criteria:

Inclusion criteria: subjects aged 19-74 belonging to the ETHON cohort or registered in the Hepamet Spanish registry of NAFLD or the European NAFLD registry

Exclusion criteria: subjects that not fulfill the inclusion criteria and those who did not sign informed consent to participate in the ETHON cohort or to be registered in the mentioned registers.

• Study Type: Observational
• Enrollment (Anticipated): 14046

Contacts and Locations

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: 15 years to 70 years (Adult, Older Adult)
• Accepts Healthy Volunteers: No
• Genders Eligible for Study: All

• Sampling Method: Non-Probability Sample

Study Population

The study has four phases: Phases I and II refer to both unsupervised and supervised artificial intelligence learning to identify clusters and build diagnostic algorithms. They will be carried out on data generated from the ETHON cohort. Phase III will consist on applying deep learning system technology as a support strategy to stratify liver biopsies in NALFD patients according to their grade of necro-inflammation and stage of fibrosis. Liver biopsies collected in the Spanish registry of NAFLD up to the beginning of the study will be used. Finally, a phase IV of validation will be performed with data from patients that are going to be registered in the European and Spanish registries of NAFLD.

Description

Inclusion Criteria:

• Subjects aged 19-74 belonging to the ETHON cohort or registered in the Hepamet Spanish registry of NAFLD or the European NAFLD registry

Exclusion Criteria:

• Subjects that not fulfill the inclusion criteria and those who did not sign informed consent to participate in the ETHON cohort or to be registered in the mentioned registers.

Study Plan

How is the study designed?

Design Details

• Observational Models: Cohort
• Time Perspectives: Other

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
ETHON; Subjects from the general population identified in the ETHON	Other: This is an observational study.; This is an observational study. No intervention is planned outside of usual clinical practice.
HEPAmet; Subjects belonging to the Spanish registry of NAFLD (HEPAmet)	Other: This is an observational study.; This is an observational study. No intervention is planned outside of usual clinical practice.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Time Frame
Number of subjects diagnosed with NAFLD and NASH in the ETHON cohort after applying Artificial Intelligence algorithms	From october of 2019 to march of 2021
Percentage of subjects diagnosed with NAFLD and NASH in the ETHON cohort after applying Artificial Intelligence algorithms	From october of 2019 to march of 2021
Sensitivity in terms of NASH diagnosis of AI algorithms with respect to histologic diagnosis compared with the Hepamet non-invasive score	From october of 2019 to march of 2021
Specificity in terms of NASH diagnosis of AI algorithms with respect to histologic diagnosis compared with the Hepamet non-invasive score	From october of 2019 to march of 2021
Positive predictive value in terms of NASH diagnosis of AI algorithms with respect to histologic diagnosis compared with the Hepamet non-invasive score.	From october of 2019 to march of 2021
Negative predictive Value in terms of NASH diagnosis of AI algorithms with respect to histologic diagnosis compared with the Hepamet non-invasive score.	From october of 2019 to march of 2021
Kappa coefficient of concordance about NASH diagnosis between AI algorithms and histologic diagnosis.	From october of 2019 to march of 2021
Kappa coefficient of concordance about NASH diagnosis between AI algorithms and the Hepamet non-invasive score.	From october of 2019 to march of 2021
ROC curve at various threshold settings obtained through the algorithms for NASH diagnosis and staging	From october of 2019 to march of 2021

Collaborators and Investigators

• Sponsor: Instituto de Investigación Marqués de Valdecilla
• Collaborators: Servicio Cántabro de Salud

Study record dates

Study Major Dates

• Study Start (Anticipated): September 30, 2019
• Primary Completion (Anticipated): September 30, 2020
• Study Completion (Anticipated): December 31, 2020

Study Registration Dates

• First Submitted: September 20, 2019
• First Submitted That Met QC Criteria: September 20, 2019
• First Posted (Actual): September 23, 2019

Study Record Updates

• Last Update Posted (Actual): September 23, 2019
• Last Update Submitted That Met QC Criteria: September 20, 2019
• Last Verified: September 1, 2019

More Information

Terms related to this study

• Keywords: NAFLD; Artificial Intelligence; non-alcoholic fatty liver disease; Deep Learning System
• Additional Relevant MeSH Terms: Digestive System Diseases; Liver Diseases; Fatty Liver; Non-alcoholic Fatty Liver Disease
• Other Study ID Numbers: NASHAI

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: No

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No