Several test sets were used, yielding 50% accuracy. The first application resulted in an accuracy of 58%. Fifty heart rate data points from both the Holter sensor and Sharp non-contact sensor were applied to the model.
The preprocessing window size for the measured heart rate data was set to 30 s, and the HRV was converted using 16 criteria. Results: After verifying the accuracies of the 12 models generated using the four algorithms and obtaining the confusion matrix of the model with the highest accuracy for each algorithm, the ensemble boosting model was selected as the final model. Learning was performed using 12 classification models, and the accuracies of the models were compared. The preprocessed open data were separated into training and test sets, and four classification model algorithms were applied for learning, those were used: decision tree, SVM, ensemble, and KNN algorithms. The MIT Arrhythmia Database of PhysioNet was used for learning, and data from a Holter monitor and Sharp non-contact sensor were used for comparison. The reliability was improved by applying a preprocessing filter algorithm and an optimal classification model. Methods: A preprocessing filter algorithm and optimal classification model were used to improve the reliability of heart rate data measured by a non-contact sensor, and the results were compared to contact sensor data. This paper presents an important advancement in heart activity monitoring, focusing on non-contact sensor data, which tend to be noisy because of the interference and limitations of its non-contact technology. It is essential that the data from non-contact sensors are as accurate as those from contact sensors. Objectives: The demand for non-contact sensors has increased owing to the COVID-19 pandemic. LA02 | Arrhythmia detection from non-contact sensor heart rate data by data transferĭepartment of Computer Engineering, Gachon University, SeongNamSi, South Korea
And, It is possible to judge the degree of recurrence to some extent in patients who have experienced colorectal cancer surgery.Īcknowledgements: This research was supported by the MSIT (Ministry of Science and ICT), Korea, under the ITRC (Information Technology Research Center) support program (IITP-2020-2017-0-01630) supervised by the IITP(Institute for Information & communications Technology Promotion).
Performance of recurrence classification model by various oversampling method.Ĭonclusions: Proposed model can effectively address data imbalance problems that arise from most medical data generated in the medical field.
0 kommentar(er)
