Beat-to-Beat Continuous Blood Pressure Estimation Using Bidirectional Long Short-Term Memory Network

Lee, Dongseok; Kwon, Hyuk‐Joon; Son, Dongyeon; Eom, Heesang; Park, Cheolsoo; Lim, Yonggyu; Seo, Chulhun; Park, Kwangsuk

doi:10.3390/s21010096

Cited by 35 publications

(12 citation statements)

References 24 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…They recruited 18 subjects and achieved an MAE of 2.62 mmHg and 2.03 mmHg for SBP and DBP after leave-one-out cross validation. After fine tuning the final dense layer of their model using 20% of their test subject’s data they were able to improve those results further [ 51 ]. A similar approach was used by Mou et al investigating different combinations of DNN, CNN and LSTM.…”

Section: Related Workmentioning

confidence: 99%

Assessment of Non-Invasive Blood Pressure Prediction from PPG and rPPG Signals Using Deep Learning

Schrumpf

Frenzel

Aust

et al. 2021

Sensors

View full text Add to dashboard Cite

Exploiting photoplethysmography signals (PPG) for non-invasive blood pressure (BP) measurement is interesting for various reasons. First, PPG can easily be measured using fingerclip sensors. Second, camera based approaches allow to derive remote PPG (rPPG) signals similar to PPG and therefore provide the opportunity for non-invasive measurements of BP. Various methods relying on machine learning techniques have recently been published. Performances are often reported as the mean average error (MAE) on the data which is problematic. This work aims to analyze the PPG- and rPPG based BP prediction error with respect to the underlying data distribution. First, we train established neural network (NN) architectures and derive an appropriate parameterization of input segments drawn from continuous PPG signals. Second, we use this parameterization to train NNs with a larger PPG dataset and carry out a systematic evaluation of the predicted blood pressure. The analysis revealed a strong systematic increase of the prediction error towards less frequent BP values across NN architectures. Moreover, we tested different train/test set split configurations which underpin the importance of a careful subject-aware dataset assignment to prevent overly optimistic results. Third, we use transfer learning to train the NNs for rPPG based BP prediction. The resulting performances are similar to the PPG-only case. Finally, we apply different personalization techniques and retrain our NNs with subject-specific data for both the PPG-only and rPPG case. Whilst the particular technique is less important, personalization reduces the prediction errors significantly.

show abstract

Section: Related Workmentioning

confidence: 99%

Assessment of Non-Invasive Blood Pressure Prediction from PPG and rPPG Signals Using Deep Learning

Schrumpf

Frenzel

Aust

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…LSTM network, which is a variant of recurrent neural network (RNN), contains cyclic feedbacks that are designed to handle the temporal sequence [25]. Thus, LSTM layers can encode relevant information of class-specific characteristics across time [26]. Owing to these characteristics, the models combined with CNN and LSTM, have been successfully applied in detecting SAHS using biosignal sequences, in recent studies [27], [28], [29].…”

Section: Introductionmentioning

confidence: 99%

Hybrid CNN-LSTM Network for Real-Time Apnea-Hypopnea Event Detection Based on IR-UWB Radar

et al. 2022

Self Cite

View full text Add to dashboard Cite

Polysomnography (PSG) is the gold-standard for sleep apnea and hypopnea syndrome (SAHS) diagnosis. Because the PSG system is not suitable for long-term continuous use owing to the high cost and discomfort caused by attached multi-channel sensors, alternative methods using a non-contact sensor have been investigated. However, the existing methods have limitations in that the radar-person distance is fixed, and the detected apnea hypopnea (AH) event cannot be provided in real-time. In this paper, therefore, we propose a novel approach for real-time AH event detection with impulse-radio ultra-wideband (IR-UWB) radar using a deep learning model. 36 PSG recordings and simultaneously measured IR-UWB radar data were used in the experiments. After the clutter was removed, IR-UWB radar images were segmented by sliding a 20-s window at 1-s shift, and categorized into two classes: AH and N. A hybrid model combining the convolutional neural networks and long short-term memory networks was trained with the data, which consisted of class-balanced segments. Time sequenced classified outputs were then fed to an event detector to identify valid AH events. Therefore, the proposed method showed a Cohen's kappa coefficient of 0.728, sensitivity of 0.781, specificity of 0.956, and an accuracy of 0.930. According to the apnea-hypopnea index (AHI) estimation analysis, the Pearson's correlation coefficient between the estimated AHI and reference AHI was 0.97. In addition, the average accuracy and kappa of SAHS diagnosis was 0.98 and 0.96, respectively, for AHI cutoffs of ≥ 5, 15, and 30 events/h. The proposed method achieved the state-of-the-art performance for classifying SAHS severity without any handengineered feature regardless of the user's location. Our approach can be utilized for a cost-effective and reliable SAHS monitoring system in a home environment.INDEX TERMS convolutional neural network, impulse-radio ultra-wideband, long short-term memory network, non-contact monitoring, sleep apnea and hypopnea syndrome

show abstract

“…In addition, recurrent neural network such as long short-term memory neural network (LSTM) [12,22] and gated recurrent unit (GRU) [14,23] have been widely employed for BP estimation, which can make predictions of current time stamp with relatively long periods of previous information, which however suffer from the high variations of input features. To combine the strengths of the two structures, CNN-LSTM or CNN-GRU based models have been proposed and become the most popular BP estimators in recent studies [14-16, 24, 25].…”

Section: Related Workmentioning

confidence: 99%

“…A practical way is to build general models based on data from a group of subjects. For general models, individual calibration is not a necessitate, depending on the availability of individual calibration data, but they usually showed lower accuracy compared to calibrated models [14,15]. However, most studies built individualized models based on the MIMIC database which recorded continuous BP as the references thus providing sufficient individual samples for training the model, i.e., normally the first 70-80% individual data were used for training the model and the remaining data for testing.…”

Section: Related Workmentioning

confidence: 99%

UTransBPNet: A General Deep Learning Model for Cuffless Blood Pressure Estimation under Activities

Gao¹,

Liu²,

Hong³

et al. 2023

Preprint

View full text Add to dashboard Cite

Cuffless blood pressure (BP) monitoring has gained great attention in the past twenty years considering its significant benefits in cardiovascular healthcare. However, the main challenge of this technology is the inaccurate BP modeling under activities, i.e., existing work have either been inappropriately validated with sufficient intra-individual BP variations, or did not show promising estimation accuracy under activities. In this study, a novel deep learning model UTransBPNet, featured in short- and long-range feature representation, is proposed aiming to improve the estimation accuracy and tracking capability of intra-individual BP changes under activities. The model performance was comprehensively evaluated in three different datasets, i.e., one public dataset (Dataset_MIMIC) and two datasets under daily activities (Dataset_Drink and Dataset_Exercise). Under subject-independent validation, the model achieved state-of-the-art performance in the Dataset_MIMIC, with the mean absolute differences (MADs) for systolic BP (SBP) and diastolic BP (DBP) of 4.38 and 2.25 mmHg, respectively. In addition, the model achieved strong tracking capability of intra-individual BP variations under activities, with the individual Pearson’ correlation coefficients for SBP and DBP of 0.61±0.17 and 0.62±0.13 (Dataset_Drink), 0.82±0.11 and 0.72±0.18 (Dataset_Exercise), respectively. Moreover, this study for the first time tested the generalization capability across different activities, and showed that with small-sized scenario-specific data for finetuning, our model showed good cross-scenario generalization capability which however degraded significantly when there are differences in the BP distribution and variation patterns between the datasets. In conclusion, UTransBPNet is generally a very promising deep learning model for accurate cuffless BP estimation and tracking BP changes. Future work should further investigate the influence of BP distribution and variation patterns on the generalization capability for building effective training dataset.

show abstract

Beat-to-Beat Continuous Blood Pressure Estimation Using Bidirectional Long Short-Term Memory Network

Cited by 35 publications

References 24 publications

Assessment of Non-Invasive Blood Pressure Prediction from PPG and rPPG Signals Using Deep Learning

Assessment of Non-Invasive Blood Pressure Prediction from PPG and rPPG Signals Using Deep Learning

Hybrid CNN-LSTM Network for Real-Time Apnea-Hypopnea Event Detection Based on IR-UWB Radar

UTransBPNet: A General Deep Learning Model for Cuffless Blood Pressure Estimation under Activities

Contact Info

Product

Resources

About