Developing Artificial Intelligence Solutions to Improve Diagnosis and Risk Stratification in Acute Pulmonary Embolism and Chronic Thrombo-Embolic Pulmonary Hypertension

Background Pulmonary Embolism (PE) is a common condition with a high mortality rate. Patients with PE are often poorly risk stratified, despite clear guidelines. Chronic Thrombo-Embolic Pulmonary Hypertension (CTEPH), a chronic complication of PE, is frequently diagnosed late, with significant morbidity and mortality impact. Guidelines advise using Right Ventricular to Left Ventricular Ratio (RV:LV) on CT Pulmonary Angiogram (CTPA) to risk stratify in acute PE, but this measurement is frequently omitted from reports. Pulmonary Artery to Aorta ratio (PA:Ao) on CTPA indicates risk of CTEPH but is similarly frequently neglected outside specialist centres. Our AI solutions automatically detect (AIDOC) and risk stratify PE (IMBIO) and CTEPH (PHAST). Aims and Objectives Clinical evaluation study Health economic model and exploratory cost-effectiveness analysis Qualitative end user research Technical scalability and adoption assessments. Hypothesis: AI solutions will improve risk stratification and thus patient outcomes. Work plan and methods Adaptations to automate output into clinical reports. Pragmatic pre/post non-randomised prospective comparative study using consecutive CTPAs 6-months pre- and 12-months post-introduction of AI to establish impact. PHAST is at an earlier stage of development: no 'live' use. Further training, validation and development. Timelines Start date May 2022. 30 month project. Anticipated Impact and Dissemination Hypotheses: Earlier identification of high risk patients who require escalation of care and low risk patients safe for ambulatory management, improving outcomes and reducing length of stay. Improved CTEPH diagnosis rate. Patient Advisory Group will aid dissemination. Our trust belongs to UK-wide PH research network; this will facilitate future multisite expansion to improve generalisability. Exit Strategy after funding Solutions are at different pipeline stages. AIDOC&IMBIO exit point: Prospective evaluation of safety, efficacy, feasibility, and cost-effectiveness of 'real-time' use in NHS. Dataset to power multicentre study. PHAST exit point: Further validation, clinical efficacy and safety data. Progression towards regulatory paperwork and commercialization strategies.

Pulmonary Embolism (PE) is a common condition that happens when blood clots cause blockages in the lung arteries. This can be a life-threatening emergency, causing death if not noticed early and treated quickly with blood-thinning medication. PE can also cause long-term problems when clots do not fully dissolve but are left behind, which can leave patients very breathless and can be fatal if the back-pressure causes the right side of the heart to fail. This long-term condition is called Chronic Thrombo-Embolic Pulmonary Hypertension (CTEPH). The sooner CTEPH is spotted and treated, the earlier potentially life-saving treatments can be started. However, there is currently a long delay to diagnosis, with many cases missed. Three key steps in PE management are: 1) prompt diagnosis, 2) assessing the level of risk, and 3) picking up long-term complications. CT scans are most commonly used for all three steps. In this study, we ll look at ways to make each step faster and more reliable, to improve outcomes for patients and reduce the time spent in hospital where possible. We will use different methods of Artificial Intelligence (AI). AI is being used more and more in the NHS to help doctors do their jobs more effectively and improve patients care. Diagnosing the PE: AIDOC . This is used to detect clots within the lung arteries and alerts clinicians as soon as a clot is detected, meaning treatment can be offered as soon as possible. Assessing the level of risk: IMBIO . This measures the widest part of both bottom chambers of the heart, the left and right ventricles. These will be calculated as a ratio , which is already known to be a useful measure to see whether there is increased pressure on the right side of the heart. Guidelines recommend that this is measured and reported in every patient with PE, to help decide whether a patient needs a special high care bed with more monitoring, or a general medical bed, or can go home – but we know that this is not reliably done currently. Long-term complications: Pulmonary Hypertension Automated Screening Technology (PHAST). This measures the width of the pulmonary artery, which carries blood into the lungs from the right side of the heart. This measurement is recommended to look for possible CTEPH, but is often not provided promptly or reliably. Scans will still be reported by human consultant radiologists (doctors who specialise in diagnosing diseases using medical imaging) and they will have final responsibility for the report but the AI technology will run in the background and ensure important findings are not missed. This will ensure clinicians have the best evidence-based information they need to help their patients in the best way they can.