Feasibility Study for Gathering Lung Sound Recordings From Pulmonary Fibrosis Patients and Healthy Controls and Classifying Disease Status Using a Combination of Digital Sound Device and Machine Learning
Overview
- Phase
- Not Applicable
- Intervention
- Not specified
- Conditions
- Pulmonary Fibrosis
- Sponsor
- Royal Devon and Exeter NHS Foundation Trust
- Enrollment
- 100
- Primary Endpoint
- Measure of ability of this system to classify participants as PF patients or healthy controls
- Status
- Not yet recruiting
- Last Updated
- 3 years ago
Overview
Brief Summary
The goal of this observational study is to test whether it is possible to detect particular lung sounds that are unique to patients with the lung disease pulmonary fibrosis and whether any such sounds could be analysed using machine learning to make diagnosing disease easier.
Participants will have a sound detection device placed in different locations on the chest and audio sounds will be recorded for analysis.
Researchers will compare audio recordings from clinically diagnosed patients with recordings from healthy controls of a similar age to see whether the sounds are sufficiently different within that age group.
Detailed Description
This is a study of chest audio recordings obtained using a sound enhancer, in this case a Bluetooth device, combined with intelligent computer-processing and analysis. It is being carried out amongst pulmonary fibrosis patients and healthy controls of a similar age, with the aim to improve diagnosis of pulmonary fibrosis and remote monitoring of disease progression. Expert respiratory doctors gain important insights about the health of a patient's lungs by listening to the chest with a stethoscope. Currently, there are insufficient respiratory experts and specialist equipment to meet the patient demand, leading to delays in diagnosis and treatment and a shortage of specialist care following diagnosis. In this study the investigators are aiming to make that specialist practice much more available by recording lung sounds and developing software to do the intelligent analysis. Initial tests with publicly available recordings of expertly diagnosed respiratory sounds have shown that different lung diseases can be detected with a very high degree of accuracy using new software. Here the investigators want to test that software with a cost-effective digital sound device in a clinical setting. The aim is for respiratory diseases to be diagnosed quickly and easily and also, in future, for patients to be offered the option to monitor how well they are after diagnosis in their own home.
Investigators
Eligibility Criteria
Inclusion Criteria
- Not provided
Exclusion Criteria
- Not provided
Outcomes
Primary Outcomes
Measure of ability of this system to classify participants as PF patients or healthy controls
Time Frame: 8 months
A measure of the capability of the machine learning model combined with the cost-effective bluetooth stethoscope to classify study participants as PF patients or healthy controls from lung sound recordings alone in a clinical setting
Feedback from patients and study clinicians
Time Frame: 8 months
Feedback from patients and study clinicians about the acceptability of digital sound monitoring for improving future diagnosis and monitoring of disease progression in pulmonary fibrosis
Number of clinical lung sound recordings stored from pulmonary fibrosis cases and controls
Time Frame: 6 months
A measure of the feasibility of gathering 12 lung sound files from each of 50 PF patients and 50 healthy volunteers in a similar age-group in the available timeframe.
Secondary Outcomes
- A correlation between clinical measures of pulmonary fibrosis severity and the audio waveform(6 months)