Application of machine learning and signal processing techniques to the design of a smart stethoscope
Lead Research Organisation:
University of Cambridge
Department Name: Engineering
Abstract
Designing machine learning and signal processing algorithms to automatically diagnose cardiovascular and respiratory illnesses, using heart and lung sound recordings obtained via an electronic stethoscope. These algorithms will then contribute towards the design of a 'smart' stethoscope that can be used by physicians to aid in patient diagnosis
People |
ORCID iD |
Anurag Agarwal (Primary Supervisor) | |
Andrew McDonald (Student) |
Publications
Thoenes M
(2021)
Narrative review of the role of artificial intelligence to improve aortic valve disease management.
in Journal of thoracic disease
Studentship Projects
Project Reference | Relationship | Related To | Start | End | Student Name |
---|---|---|---|---|---|
EP/N509620/1 | 30/09/2016 | 29/09/2022 | |||
1936431 | Studentship | EP/N509620/1 | 30/09/2017 | 29/09/2021 | Andrew McDonald |
Description | As part of this award, machine learning algorithms to automatically detect abnormal heart murmurs and screen for clinically significant valvular heart disease have been developed. To deploy these algorithms, we have developed a proof-of-concept device that pairs an off-the-shelf electronic stethoscope with a smartphone application to form a screening tool for valve disease. Key to the development of these machine learning algorithms has been the collection of patient data. Heart sound data is extremely limited because electronic stethoscopes are not widespread and heart sounds are not routinely recorded as part of the patient record. As part of this award, an NHS study to collect heart sound data has been designed and is currently operating in multiple trusts across the UK. The resulting curated data from NHS patients is an excellent resource for future training of machine learning algorithms for valve disease diagnosis. |
Exploitation Route | The technology developed as part of this award will be commercialised as a spin-out from the University of Cambridge, in collaboration with Cambridge Enterprise. It can be used as a screening tool for valvular heart disease, helping to detect asymptomatic cases earlier and therefore improve patient prognoses and reduce burden on hospital care. |
Sectors | Digital/Communication/Information Technologies (including Software) Healthcare |
Description | Currently, the proof-of-concept VHD screening tool has been used as a demonstrator, to raise awareness of valvular heart disease and possible improvements to the patient care pathway. This has included at events such as a Parliament MedTech reception, a screening event in Birmingham, and a roundtable valve disease discussion, all in collaboration with leading charity Heart Valve Voice. |
First Year Of Impact | 2020 |
Sector | Healthcare |
Impact Types | Societal Policy & public services |
Description | Innovation in Heart Valve Disease Meeting |
Geographic Reach | Local/Municipal/Regional |
Policy Influence Type | Participation in a guidance/advisory committee |
Description | An Intelligent Stethoscope |
Amount | £651,538 (GBP) |
Funding ID | MR/S036644/1 |
Organisation | Medical Research Council (MRC) |
Sector | Public |
Country | United Kingdom |
Start | 05/2019 |
End | 02/2023 |
Description | Kennel Club Charitable Trust Scientific Grant |
Amount | £17,800 (GBP) |
Organisation | The Kennel Club Charitable Trust |
Sector | Charity/Non Profit |
Country | United Kingdom |
Start | 01/2019 |
End | 01/2021 |
Description | Wellcome Trust Access to Expertise grant |
Amount | £15,000 (GBP) |
Organisation | Wellcome Trust |
Sector | Charity/Non Profit |
Country | United Kingdom |
Start | 04/2021 |
End | 10/2021 |
Title | Canine valvular heart disease dataset |
Description | Electronic stethoscope recordings from dogs attending hospital referral clinics, with heart murmur assessment performed by a cardiologist and matching echocardiogram diagnoses of valvular heart disease. To date 189 dogs and 759 heart sound recordings have been made. |
Type Of Material | Database/Collection of data |
Year Produced | 2021 |
Provided To Others? | No |
Impact | Ongoing development of a canine heart murmur detection algorithm. |
Title | Equine valvular heart disease dataset |
Description | Electronic stethoscope recordings, cardiologist heart murmur assessments and echocardiogram diagnoses from horses attending a veterinary surgeons. |
Type Of Material | Database/Collection of data |
Year Produced | 2020 |
Provided To Others? | No |
Impact | Ongoing development of equine murmur assessment machine learning algorithm. |
Title | Valvular heart disease dataset for development of intelligent stethoscope |
Description | Collation of heart sound stethoscope recordings, echocardiogram images and anonymised patient metadata. For use in training machine learning algorithms to predict valvular heart disease from stethoscope data. Diagnosis of valvular heart disease is done by echocardiogram and used as the ground truth labels. |
Type Of Material | Database/Collection of data |
Year Produced | 2020 |
Provided To Others? | No |
Impact | Ongoing development of machine learning algorithm to predict significant valvular heart disease from stethoscope recordings. |
Description | Cambridge Enterprise |
Organisation | University of Cambridge |
Department | Cambridge Enterprise |
Country | United Kingdom |
Sector | Academic/University |
PI Contribution | Development of technology and commercial strategy to bring valvular heart disease screening tool to market. |
Collaborator Contribution | Assisting with all aspects of commercialisation of the technology, including follow-on funding, design of business plan, and networking with mentors and potential interested parties. |
Impact | A plan for a spin-out from the university, BioPhonics. |
Start Year | 2017 |
Description | Collaboration with Papworth Hospital for clinical studies |
Organisation | Royal Papworth Hospital NHS Foundation Trust |
Country | United Kingdom |
Sector | Public |
PI Contribution | Application for funding from MRC for study titled "Cardiovascular Acoustics and an Intelligent Stethoscope". Design and management of study, which is collecting heart sounds and echocardiograms from valvular heart disease patients in the NHS. Research and development of an algorithm for automated diagnosis of valvular heart disease from the heart sounds collected. |
Collaborator Contribution | Co-applicant for funding from MRC. Design and on-going management of the study, and curation of the resulting dataset. Providing clinical knowledge to aid design of algorithms. |
Impact | Ongoing study "Cardiovascular Acoustics and an Intelligent Stethoscope" into valvular heart disease and AI diagnosis. |
Start Year | 2015 |
Description | Veterinary cardiologists at Queen's Veterinary School Hospital |
Organisation | University of Cambridge |
Department | Veterinary School |
Country | United Kingdom |
Sector | Academic/University |
PI Contribution | Development of machine learning algorithms to detect murmurs in canine heart sound data Curation of datasets |
Collaborator Contribution | Collection of patient data, management of study, and contribution of expertise on canine valvular heart disease. |
Impact | Ongoing development and validation of AI algorithm to detect murmurs in canine heart sound data |
Start Year | 2018 |
Title | Smart stethoscopes |
Description | Development of machine learning model for automatic detection of murmurs in stethoscope recordings |
IP Reference | WO2019GB50461 |
Protection | Patent application published |
Year Protection Granted | |
Licensed | No |
Impact | n/a |
Title | Cardiovascular Acoustics and an Intelligent Stethoscope Research Study |
Description | An observational research study to collect heart sound recordings from patients to understand the acoustic characteristics of murmurs and develop an artificially intelligent stethoscope. The research study is collating a dataset of stethoscope heart sound recordings, echocardiogram ultrasound images and patient data, which are being used to train machine learning algorithms to predict valvular heart disease from heart sounds. The study is funded by a Development Pathway Funding Scheme from the Medical Research Council (MR/S036644/1). |
Type | Diagnostic Tool - Non-Imaging |
Current Stage Of Development | Refinement. Clinical |
Year Development Stage Completed | 2020 |
Development Status | Under active development/distribution |
Impact | Still under development. |
URL | https://www.hra.nhs.uk/planning-and-improving-research/application-summaries/research-summaries/card... |
Company Name | BioPhonics |
Description | BioPhonics develops a stethoscope that utilises AI technology to detect murmurs more effectively. |
Year Established | 2018 |
Impact | n/a |
Website | https://biophonics.co.uk/ |
Description | Innovation in Valve Disease Meeting |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Professional Practitioners |
Results and Impact | Meeting of NHS consultants, industry professionals, and academics, to discuss innovative methods to improve patient care across the heart valve disease pathway. We presented a demo of our automated heart murmur detection software and discussed how it could be implemented in a practical setting. The discussion raised several interesting points for future research and led to further opportunities for collaboration with clinicians and policymakers. |
Year(s) Of Engagement Activity | 2020 |
URL | http://healthinnovationmanchester.com/dr-tracey-vell-mbe-innovation-in-valve-disease-implementing-te... |
Description | Medtronic "Your Heart Matters" Bus valve disease screening event |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | Local |
Primary Audience | Public/other audiences |
Results and Impact | Demonstrated our valve disease technology to general public, clinicians and other stakeholders at a valve disease screening event held in central Birmingham. |
Year(s) Of Engagement Activity | 2022 |
Description | Parliament MedTech Awareness Week reception |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Policymakers/politicians |
Results and Impact | Attended Parliament MedTech Awareness Week reception, where we demonstrated our proof-of-concept valve disease screening tool to attendees. Helped to raise awareness of valve disease and show a possible solution to detect patients earlier. |
Year(s) Of Engagement Activity | 2021 |
URL | https://mtg.org.uk/2021-2/ |