Artificial intelligence to improve the detection and risk stratification of acute pulmonary embolism (AID-PE): protocol for a pragmatic quasi-experimental comparator study.
Authors
Affiliations (12)
Affiliations (12)
- Department of Research and Development, Royal United Hospitals Bath NHS Foundation Trust, Bath, UK.
- Department of Respiratory, Royal United Hospitals Bath NHS Foundation Trust, Bath, UK.
- Department of Radiology, Royal United Hospitals Bath NHS Foundation Trust, Bath, UK.
- Department of Medical Statistics, London School of Hygiene and Tropical Medicine, London, UK.
- Department for Health, University of Bath, Bath, UK.
- The National Institute for Health and Care Research Applied Research Collaboration West (NIHR ARC West), University Hospitals Bristol and Weston NHS Foundation Trust, Bristol, UK.
- Health Economics and Health Policy at Bristol (HEHP@Bristol), University of Bristol, Bristol, UK.
- Bath Pulmonary Hypertension Patient Advisory Group, Royal United Hospitals Bath NHS Foundation Trust, Bath, UK.
- Department of Mechanical Engineering, University of Bath, Bath, England, UK.
- Department of Life Sciences, University of Bath, Bath, UK.
- Department of Radiology, Royal United Hospitals Bath NHS Foundation Trust, Bath, UK [email protected].
- Department of Health, University of Bath, Bath, UK.
Abstract
Pulmonary embolism (PE) is a potentially fatal condition requiring timely diagnosis and treatment. CT pulmonary angiography (CTPA) is the gold standard for diagnosis and indicates PE severity through radiological markers of right heart strain. However, accurate interpretation and communication of these findings is often suboptimal in real-world practice. Artificial intelligence (AI) could alleviate pressure on radiology services by supporting PE identification, risk stratification and worklist prioritisation. Before widespread adoption, AI tools must be rigorously validated for diagnostic accuracy, safety and clinical impact. This pragmatic single-centre, non-randomised quasi-experimental study will evaluate the diagnostic accuracy, feasibility, and clinical-cost impact of AI-assisted PE detection and risk stratification using AIDOC and IMBIO software. We will recruit two consecutive cohorts of adult patients undergoing CTPAs for suspected PE: a comparator cohort (12 months pre-AI implementation) and an intervention cohort (12 months post-AI implementation). AI will be applied retrospectively to the comparator cohort, while in the intervention cohort, radiologists will have contemporaneous access to the AI's interpretation of CTPA images.A subset of retrospective scans, both PE-positive and PE-negative, will undergo expert thoracic radiologist review to establish a reference standard. Data on patient demographics, clinical management and outcomes will be collected. Clinical management pathways and patient outcomes will be compared between cohorts to assess AI's influence on acute PE management. Health economic modelling will assess the cost-effectiveness of integrating AI technology within the diagnostic workflow of acute PE. This study was approved by the UK Healthcare Research authority (IRAS 311735, 10 May 2023). Ethical approval was granted by West of Scotland Research Ethics Service (23/WS/0067, 3 May 2023). Results will be shared with stakeholders, presented at national and international conferences, and published in open-access peer-reviewed journals. NCT06093217.