Evaluation of AI-Assisted Detection of Malignant Lung Nodules in Routine Chest X-Rays

Background

Lung cancer is India's fastest-growing cancer burden, with cases projected to nearly double between 2015 and 2025. Outcomes are steeply stage-dependent: Survival exceeds 70% when caught early but falls below 10% at advanced stages and yet, most cases are diagnosed late. Standard screening (low-dose CT) targets only older, high-risk smokers, missing the substantial share of Indian lung cancers driven by air pollution, biomass fuel exposure, and genetic factors in never-smokers.

Study Focus Areas

This study evaluates whether exposing medical officers to Qure.ai's AI-generated lung nodule malignancy risk score (LNMS) changes clinical decision-making behavior by shifting the threshold of suspicion for malignancy. It examines whether the AI-assisted pathway increases CT referrals and investigates the reasoning behind referral decisions including how a clinician's experience, specialization, and confidence in CXR interpretation shape whether a flagged case is acted on, as well as the clinical subtypes of cases that are being acted on.

Methods

This is a real-world implementation study using a pre-post evaluation that compares standard practice to AI-supported diagnostic pathways within the same facilities. A nested behavioral study also examines how clinicians reason through and act on AI-assisted risk information in routine clinical encounters.

Intended Outcome

Existing evidence for AI-assisted nodule scoring comes from controlled, research-driven studies that cannot show whether the tool changes real clinician behavior or detection rates in routine practice. This study fills that gap by generating real-world evidence from generalist medical officers working without radiology support, in an existing resource-constrained referral pathway. Findings will clarify not just whether the tool works, but how and for whom it adds value, informing responsible scale-up decisions by the Telangana state health system and comparable secondary-care settings across India and other low- and middle-income countries.