Multimodal Cardiomegaly Classification with Image-Derived Digital Biomarkers

Duvieusart, Benjamin; Krones, Felix; Parsons, Guy; Tarassenko, Lionel; Papież, Bartłomiej W.; Mahdi, Adam

doi:10.1007/978-3-031-12053-4_2

Cited by 3 publications

(3 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A direction of future research is developing a superior method for integrating patients' demographic data and signal features with the outputs from DBRes. This could include adjusting the objective function of XGBoost to align with the challenge score or investigating other methods of multimodal data fusion [12].…”

Section: Discussionmentioning

confidence: 99%

Dual Bayesian ResNet: A Deep Learning Approach to Heart Murmur Detection

Krones¹,

Walker²,

Mahdi³

et al. 2022

Computing in Cardiology Conference (CinC)

View full text Add to dashboard Cite

This study presents our team PathToMyHeart's contribution to the George B. Moody PhysioNet Challenge 2022. Two models are implemented. The first model is a Dual Bayesian ResNet (DBRes), where each patient's recording is segmented into overlapping log mel spectrograms. These undergo two binary classifications: present versus unknown or absent, and unknown versus present or absent. The classifications are aggregated to give a patient's final classification. The second model is the output of DBRes integrated with demographic data and signal features using XGBoost. DBRes achieved our best weighted accuracy of 0.771 on the hidden test set for murmur classification, which placed us fourth for the murmur task. (On the clinical outcome task, which we neglected, we scored 17th with costs of 12637.) On our held-out subset of the training set, integrating the demographic data and signal features improved DBRes's accuracy from 0.762 to 0.820. However, this decreased DBRes's weighted accuracy from 0.780 to 0.749. Our results demonstrate that log mel spectrograms are an effective representation of heart sound recordings, Bayesian networks provide strong supervised classification performance, and treating the ternary classification as two binary classifications increases performance on the weighted accuracy.

show abstract

Section: Discussionmentioning

confidence: 99%

Dual Bayesian ResNet: A Deep Learning Approach to Heart Murmur Detection

Krones¹,

Walker²,

Mahdi³

et al. 2022

Computing in Cardiology Conference (CinC)

View full text Add to dashboard Cite

show abstract

“…Alternative methodologies based on robust feature engineering, such as segmentation, have been explored by other leading teams [83]. Such approaches of robust feature engineering offer potential improvements in interpretability and may enhance model robustness against overfitting [84].…”

Section: Limitations and Future Modelling Researchmentioning

confidence: 99%

“…Additionally, the exploration of multiple fusion techniques may improve model performance [85]. We fused the features at a relatively late stage; however, an earlier feature fusion could better align with how clinicians use demographic information when interpreting charts [84]. Besides multimodal approaches, there are many other possible paths to explore to increase model robustness.…”

Section: Limitations and Future Modelling Researchmentioning

confidence: 99%

From Theoretical Models to Practical Deployment: A Perspective and Case Study of Opportunities and Challenges in AI-driven Cardiac Auscultation Research for Low-Income Settings

Krones,

Walker

2023

Preprint

Self Cite

View full text Add to dashboard Cite

This paper critically explores the opportunities and challenges of deploying Artificial Intelligence (AI) in healthcare. This study has two parallel components: (1) A narrative literature summary, which assesses the capacity of AI to aid in addressing the observed disparity in healthcare between high- and low-income countries. Despite the development of machine learning models for a wide range of diseases, many are never deployed in practice. We highlight various challenges that contribute to the lack of deployed models. A main challenge that is not always sufficiently addressed in the literature is the evaluation of model generalisation. For example, by using a multi-site set-up with test sets that were collected separately to the train and validation sets, or by using evaluation metrics which are both understandable and clinically applicable. Moreover, we discuss how the emerging trend of human-centred deployment research is a promising avenue for overcoming barriers towards deployment. (2) A case study on developing and evaluating a predictive AI model tailored for low-income environments. The focus of this case study is heart murmur detection in rural Brazil. Our Binary Bayesian ResNet model leverages overlapping log mel spectrograms of patient heart sound recordings and integrates demographic data and signal features via XGBoost to optimise performance. We discuss the model’s limitations, its robustness, and the obstacles preventing its practical application. We especially highlight how our model, and other state-of-the-art models, struggle to generalise to out-of-distribution data. The research accentuates the transformative potential of AI-enabled healthcare, particularly affordable point-of-care monitoring systems, in low-income settings. It also emphasises the necessity for effective implementation and integration strategies to guarantee the successful deployment of these technologies.

show abstract

From theoretical models to practical deployment: A perspective and case study of opportunities and challenges in AI-driven cardiac auscultation research for low-income settings

Krones,

Walker

2024

PLOS Digit Health

View full text Add to dashboard Cite

This article includes a literature review and a case study of artificial intelligence (AI) heart murmur detection models to analyse the opportunities and challenges in deploying AI in cardiovascular healthcare in low- or medium-income countries (LMICs). This study has two parallel components: (1) The literature review assesses the capacity of AI to aid in addressing the observed disparity in healthcare between high- and low-income countries. Reasons for the limited deployment of machine learning models are discussed, as well as model generalisation. Moreover, the literature review discusses how emerging human-centred deployment research is a promising avenue for overcoming deployment barriers. (2) A predictive AI screening model is developed and tested in a case study on heart murmur detection in rural Brazil. Our binary Bayesian ResNet model leverages overlapping log mel spectrograms of patient heart sound recordings and integrates demographic data and signal features via XGBoost to optimise performance. This is followed by a discussion of the model’s limitations, its robustness, and the obstacles preventing its practical application. The difficulty with which this model, and other state-of-the-art models, generalise to out-of-distribution data is also discussed. By integrating the results of the case study with those of the literature review, the NASSS framework was applied to evaluate the key challenges in deploying AI-supported heart murmur detection in low-income settings. The research accentuates the transformative potential of AI-enabled healthcare, particularly for affordable point-of-care screening systems in low-income settings. It also emphasises the necessity of effective implementation and integration strategies to guarantee the successful deployment of these technologies.

show abstract

Multimodal Cardiomegaly Classification with Image-Derived Digital Biomarkers

Cited by 3 publications

References 31 publications

Dual Bayesian ResNet: A Deep Learning Approach to Heart Murmur Detection

Dual Bayesian ResNet: A Deep Learning Approach to Heart Murmur Detection

From Theoretical Models to Practical Deployment: A Perspective and Case Study of Opportunities and Challenges in AI-driven Cardiac Auscultation Research for Low-Income Settings

From theoretical models to practical deployment: A perspective and case study of opportunities and challenges in AI-driven cardiac auscultation research for low-income settings

Contact Info

Product

Resources

About