The early diagnosis of Interstitial Lung Disease (ILD) is challenging for clinicians as lung function is only minimally affected at the onset. To improve early diagnosis, this study aims to explore the potential of AI software in assisting pulmonologists with ILD diagnosis. It provides an automated description of lung function in accordance with the most recent ATS/ERS guidelines and used an AI model to compute disease probabilities.

In Phase 1, 60 patients, 30 of whom had ILD, were retrospectively diagnosed by 25 pulmonologists by evaluating a full Pulmonary Function Test (body plethysmography, diffusion) and a short anamnesis. The experts screened the cohort twice, first without the aid of AI, and then later with the assistance of ArtiQ.PFT (v1.4, ArtiQ, BE) software The pulmonologists provided a primary diagnosis and up to three differential diagnoses for each case. In Phase 2 of the study, half of the experts repeated the protocol after using ArtiQ.PFT for six months.

In Phase 1, AI improved the detection of ILD as the primary diagnosis from 42.8% without AI to 72.1% with AI (+68.5%). A similar conclusion holds considering the differential diagnoses, with AI improving ILD diagnosis from 60.0% to 86.8% (+39.6%). Experts did not change the number of diagnoses given (2.2±0.9) with or without AI. Phase 2 yielded a similar outcome: using AI increased ILD diagnosis based on primary diagnosis (52.5% to 77.9%) and based on all the diagnoses (64.7% to 90.3%). The increase in accuracy between the phases indicates that pulmonologists improved through the use of ArtiQ.PFT.

This study shows that AI-based decision support on PFT interpretation improves accurate and early ILD diagnosis.