July 31, 2017Physiological MeasurementOpen Access

DropConnected neural networks trained on time-frequency and inter-beat features for classifying heart sounds

View Full Paper

Ask AI

Bookmark

Population

Heart sounds from the PhysioNet open-access heart sounds database

Design

Other

Authors

EKEdmund KayUniversity of Nottingham Ningbo China AAAnurag AgarwalBetsi Cadwaladr University Health Board

Discussion

Member takes

Implication

May enable automated valvular screening in primary care or low-resource settings; leaves open prospective validation before clinical use.

Structured PICO

Population

Heart sounds from the PhysioNet open-access heart sounds database

Intervention

DropConnected neural networks trained on time-frequency and inter-beat features

Outcome

Accuracy of classifying heart sounds as normal or abnormalsurrogate

A neural network algorithm using time-frequency and inter-beat features can classify heart sounds with moderate realistic accuracy, highlighting the potential for automated auscultation in primary care.

Limitations

Test data contained a dataset (dataset-e) in which normal and abnormal heart sounds were recorded with different stethoscopes, making the initial test accuracy unrealistic of real-world performance.

Cite This Study

Kay et al. (2017) studied this question.

synapsesocial.com/papers/6a72e4c4c632065ce960d50a https://doi.org/10.1088/1361-6579/aa6a3d

View Full Paper

Ask AI

Bookmark

DropConnected neural networks trained on time-frequency and inter-beat features for classifying heart sounds

Discussion

Member takes

Implication

Structured PICO

Limitations

Cite This Study

DropConnected neural networks trained on time-frequency and inter-beat features for classifying heart sounds

Discussion

Member takes

Implication

Structured PICO

Limitations

Cite This Study

Also Consider

Also Consider