Detection of subclinical rheumatic heart disease in children using a deep learning algorithm on digital stethoscope: a study protocol

Abstract

IntroductionRheumatic heart diseases (RHDs) contribute significant morbidity and mortality globally. To reduce the burden of RHD, timely initiation of secondary prophylaxis is important. The objectives of this study are to determine the frequency of subclinical RHD and to train a deep learning (DL) algorithm using waveform data from the digital auscultatory stethoscope (DAS) in predicting subclinical RHD.Methods and analysisWe aim to recruit 1700 children from a group of schools serving the underprivileged over a 12-month period in Karachi (Pakistan). All consenting students within the age of 5–15 years with no underlying congenital heart disease will be eligible for the study. We will gather information regarding sociodemographics, anthropometric data, history of symptoms or diagnosis of rheumatic fever, phonocardiogram (PCG) and electrocardiography (ECG) data obtained from DAS. Handheld echocardiogram will be performed on each study participant to assess the presence of a mitral regurgitation (MR) jet (>1.5 cm), or the presence of aortic regurgitation (AR) in any view. If any of these findings are present, a confirmatory standard echocardiogram using the World Heart Federation (WHF) will be performed to confirm the diagnosis of subclinical RHD. The auscultatory data from digital stethoscope will be used to train the deep neural network for the automatic identification of patients with subclinical RHD. The proposed neural network will be trained in a supervised manner using labels from standard echocardiogram of the participants. Once trained, the neural network will be able to automatically classify the DAS data in one of the three major categories—patient with definite RHD, patient with borderline RHD and normal subject. The significance of the results will be confirmed by standard statistical methods for hypothesis testing.Ethics and disseminationEthics approval has been taken from the Aga Khan University, Pakistan. Findings will be disseminated through scientific publications and to collaborators.Article focusThis study focuses on determining the frequency of subclinical RHD in school-going children in Karachi, Pakistan and developing a DL algorithm to screen for this condition using a digital stethoscope.

Machine learning methods for predicting major types of rheumatic heart diseases in children of Southern Punjab, Pakistan

“…MS was the second most common lesion of which 7.0% had mild, 8.1% had moderate, and 55.8% had severe ( 13 ). The reported prevalence of RHD in Pakistan was found to be 22/1,000 in school-going children ( 14 ). Studies in Bangladesh, India, and Pakistan showed that the common non-clinical predictors of RHD were age, gender, living place, family size, number of people sharing a living room, wall material used in the home, number of siblings, and maternal occupation ( 4 , 15 , 16 ).…”

Artificial Intelligence in Pediatric Cardiology: A Scoping Review

“…The limitations of objective performance data have restricted wide acceptability so far [ 90 , 91 , 92 ]. The auscultatory findings of CHD are integral to the clinical diagnosis and have the benefit of being a low-cost tool, but are subject to clinical expertise, which becomes a limitation in resource-limited countries and creates a need for support to the clinician [ 93 , 94 , 95 , 96 ] that is objective and reportable by even peripheral health workers. The lack of trained cardiologists at the peripheral level leads to an unavoidable miss of a timely clinical diagnosis of CHD, contributing to delayed intervention and thus a poor prognosis [ 97 ].…”

Recent Advances in the Rheumatic Fever and Rheumatic Heart Disease Continuum

“…Further developments toward a more pragmatic uptake of echocardiographic screening in LMICs have included: (a) development of hand-held echo devices which are less expensive, generate smaller and shareable file sizes, and have a smaller profile without reliance on wired electricity [51]; (b) task-shifting of echo image acquisition, from highly trained cardiologists and echocardiographers to non-expert health care workers (HCW) with limited training, allowing for expansion of screening capabilities; (c) supportive cloud-based telemedicine programs for remote echo image interpretation; and finally, (d) expanding technological advances [54,55], with several companies now offering artificial intelligence guidance technology built into hand-held devices, thus empowering novice operators to perform screening echoes even after only minimal training, and promising deep learning approaches for automated flagging of abnormalities in the near future [56]. There is also ongoing effort to improve and simplify the WHF criteria for the diagnosis of latent RHD [57], with additional development of risk scores to predict outcomes in both borderline and definite latent RHD [57].…”