A systematic review of artificial intelligence-based diagnosis models for idiopathic pulmonary fibrosis

Idiopathic pulmonary fibrosis (IPF) is a major and difficult disease with unknown etiology and continuous progression, for which early and accurate diagnosis is challenging. With the advancement of medical technology and science, artificial intelligence (AI) has achieved significant results in IPF diagnosis and prediction of patient prognosis. However, the diagnostic and predictive performance of these AI models still lacks comprehensive evidence. Therefore, this study systematically reviewed and critically appraised the diagnostic performance of the model in IPF patients, aiming to promote the development of future related research.

A computerized systematic search was conducted in the China National Knowledge Infrastructure (CNKI), Wanfang, China Science and Technology Journal Database (VIP), PubMed, The Cochrane Library, Web of Science and Embase for relevant literature on IPF diagnosis models, with the search period ranging from database inception to December 1, 2025. Three researchers independently screened the literature. Data were extracted according to the key assessment and data extraction Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS). The risk of bias and applicability of the models were assessed using the Prediction Model Risk of Bias Assessment Tool (PROBAST). The quality of model reporting was evaluated using the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) checklist.

A total of 11 studies were included, all of which reported on the development and validation of the model. The most common predictive factor included in the model is genes. In terms of bias risk, 3 studies were rated as high bias risk, with bias risk mainly coming from outcome reporting and analysis areas. In terms of applicability, 4 studies were rated as high-risk and 7 studies were rated as unclear, indicating low clinical applicability of the model.

Currently, the predictive model for IPF diagnosis is still in the exploratory stage, with good model discrimination but high overall risk of bias. In the future, research design should be optimized and the reporting process should be improved to ensure the development of clinically practical predictive models.