Pulmonary Hypertension (Diagnosis)
Conditions
Keywords
Artificial Intelligence, Machine Learning, Electrocardiogram, convolutional neural network
Brief summary
The goal of this observational study is to apply Artificial Intelligence (AI) and machine learning technology to the resting 12-lead electrocardiogram (ECG) and assess whether it can assist doctors in the early diagnosis of Pulmonary Hypertension (PH). Early and accurate diagnosis is an important step for patients with PH. It helps provide effective treatments early which improve prognosis and quality of life. The main questions our study aims to answer are: 1. Can AI technology in the 12-lead ECG accurately predict the presence of PH? 2. Can AI technology in the 12-lead ECG identify specific sub-types of PH? 3. Can AI technology in the 12-lead ECG predict mortality in patients with PH? In this study, the investigators will recruit 12-lead ECGs from consenting participants who have undergone Right heart Catheterisation (RHC) as part of their routine clinical care. AI technology will be applied to these ECGs to assess whether automated technology can predict the presence of PH and it's associated sub-types.
Detailed description
This study will be led by Royal United Hospital Bath NHS Trust and Liverpool John Moore's University. The aim of this study is to utilise Artificial Intelligence (AI) and machine learning technology to assist clinicians in the early diagnosis of Pulmonary Hypertension (PH). We hypothesise that the AI technologies can improve the quantification and interpretation of the parameters involved in detecting PH. This is either through highlighting significant abnormalities in the 12-lead ECG, or by rapidly providing fully automated measures of the features on the 12-lead ECG which indicate PH. The combination of these electrocardiographic features with clinical data may provide highly accurate predictive tools. This observational study will have a retrospective and prospective arm with a 3 year follow-up period. Participants will not require any additional tests or procedures at any point during the study. Any ECGs performed within the 12 months prior to a participant's right heart catheterisation (RHC) will undergo Artificial Intelligence analysis to establish if early indicators of PH are identifiable. For all recruited participants, an anonymised clinician case report form will be used to capture details relating to their demographics and routine clinical care. Follow-up times and outcomes including mortality and morbidity will also be recorded.
Interventions
DIAGNOSTIC_TESTArtificial Intelligence and Machine Learning technology
Artificial Intelligence describes computer software designed to mimic human cognitive function. Machine learning is a type of artificial intelligence in which the model created is exposed to data, identifies patterns, and recognises relationships between features seen in the data and the 'ground truth'. This technology will be applied to participants ECGs.
Sponsors
Liverpool John Moores University
Royal United Hospitals Bath NHS Foundation Trust
Study design
Observational model
COHORT
Time perspective
OTHER
Eligibility
Sex/Gender
ALL
Age
18 Years to No maximum
Healthy volunteers
Yes
Inclusion criteria
1. prospective cohort: From July 2023, all patients aged 18 or over who are referred to the Bath Pulmonary Hypertension shared care service with clinical suspicion of PH and, who through their routine clinical care, undergo a RHC and 12-lead ECG. 2. Retrospective cohort: All patients aged 18 or over who were referred to the local Pulmonary Hypertension shared care service between 2007 and June 2023, and through their routine clinical care, have undergone RHC within a year of a 12-lead ECG. This cohort will also include patients who are deceased.
Exclusion criteria
* Patient's less than 18 years-old * Patients who do not give valid consent (except deceased patients; REC approved) * Patients who have not undergone RHC to assess for PH * Patients who have not had an ECG within 12 months of their RHC
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Pulmonary Hypertension diagnosis | baseline | The investigators will calculate the area under the receiver operating characteristic curve (AUROC) for PH diagnosis by artificial intelligence technology and compare this to RHC (the gold standard) |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Pulmonary Hypertension sub-type | baseline | The investigators will assess the diagnostic test accuracy of Artificial Intelligence technology to categorise participant ECGs according to Pulmonary Hypertension sub-type and compare this to standard clinical assessment |
| Mortality | 3 years | The investigators will calculate the area under the receiver operating characteristics curve (AUROC) for mortality as predicted by Artificial Intelligence technology |
| Morbidity | baseline | The investigators will calculate the area under the receiver operating characteristics curve for morbidity as predicted by Artificial Intelligence technology and compare this to current measures (NYHA functional class, 6MWT, Pulmonary function tests) |
Countries
United Kingdom
Outcome results
None listed