DeepHeart: Semi-Supervised Sequence Learning for Cardiovascular Risk Prediction

Ballinger, Brandon; Hsieh, Johnson; Singh, Avesh; Sohoni, Nimit S.; Wang, Jack; Tison, Geoffrey H.; Marcus, Gregory M.; Sánchez, José M.; Maguire, Carol; Olgin, Jeffrey E.; Pletcher, Mark J.

doi:10.1609/aaai.v32i1.11891

Cited by 83 publications

(24 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This conceptual framework, although demonstrated here with the Catch22 method [ 39 ], is agnostic to the choice of the feature representation method for time series data [ 36 , 37 ]. Furthermore, in contrast to black-box feature learning methods based on large labeled data sets [ 31 ], our approach yields more interpretable time series features with smaller unlabeled data sets.…”

Section: Discussionmentioning

confidence: 99%

“…However, the analysis of time series data recorded in free-living states is challenging, as these data tend to exhibit real-world noise and fluctuations and typically lack important physical and physiological contexts. A few recent studies have used black-box deep neural networks to relate high-resolution heart rate and step count time series recorded using wearables to the risk of developing atrial fibrillation, sleep apnea, and hypertension [ 31 , 32 ]. As their primary goal focused on risk target classification, the nature of the intermediate predictive time series features and their connection with known clinical and biological markers of cardiometabolic disease remains unresolved.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

High-Resolution Digital Phenotypes From Consumer Wearables and Their Applications in Machine Learning of Cardiometabolic Risk Markers: Cohort Study

Zhou¹,

Chan²,

Foo³

et al. 2022

J Med Internet Res

View full text Add to dashboard Cite

Background Consumer-grade wearable devices enable detailed recordings of heart rate and step counts in free-living conditions. Recent studies have shown that summary statistics from these wearable recordings have potential uses for longitudinal monitoring of health and disease states. However, the relationship between higher resolution physiological dynamics from wearables and known markers of health and disease remains largely uncharacterized. Objective We aimed to derive high-resolution digital phenotypes from observational wearable recordings and to examine their associations with modifiable and inherent markers of cardiometabolic disease risk. Methods We introduced a principled framework to extract interpretable high-resolution phenotypes from wearable data recorded in free-living conditions. The proposed framework standardizes the handling of data irregularities; encodes contextual information regarding the underlying physiological state at any given time; and generates a set of 66 minimally redundant features across active, sedentary, and sleep states. We applied our approach to a multimodal data set, from the SingHEART study (NCT02791152), which comprises heart rate and step count time series from wearables, clinical screening profiles, and whole genome sequences from 692 healthy volunteers. We used machine learning to model nonlinear relationships between the high-resolution phenotypes on the one hand and clinical or genomic risk markers for blood pressure, lipid, weight and sugar abnormalities on the other. For each risk type, we performed model comparisons based on Brier scores to assess the predictive value of high-resolution features over and beyond typical baselines. We also qualitatively characterized the wearable phenotypes for participants who had actualized clinical events. Results We found that the high-resolution features have higher predictive value than typical baselines for clinical markers of cardiometabolic disease risk: the best models based on high-resolution features had 17.9% and 7.36% improvement in Brier score over baselines based on age and gender and resting heart rate, respectively (P<.001 in each case). Furthermore, heart rate dynamics from different activity states contain distinct information (maximum absolute correlation coefficient of 0.15). Heart rate dynamics in sedentary states are most predictive of lipid abnormalities and obesity, whereas patterns in active states are most predictive of blood pressure abnormalities (P<.001). Moreover, in comparison with standard measures, higher resolution patterns in wearable heart rate recordings are better able to represent subtle physiological dynamics related to genomic risk for cardiometabolic disease (improvement of 11.9%-22.0% in Brier scores; P<.001). Finally, illustrative case studies reveal connections between these high-resolution phenotypes and actualized clinical events, even for borderline profiles lacking apparent cardiometabolic risk markers. Conclusions High-resolution digital phenotypes recorded by consumer wearables in free-living states have the potential to enhance the prediction of cardiometabolic disease risk and could enable more proactive and personalized health management.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

High-Resolution Digital Phenotypes From Consumer Wearables and Their Applications in Machine Learning of Cardiometabolic Risk Markers: Cohort Study

Zhou¹,

Chan²,

Foo³

et al. 2022

J Med Internet Res

View full text Add to dashboard Cite

show abstract

“…This abundance of sensory data has kickstarted the development of several applications focused on general user and health monitoring, as well as other predictive analytic tasks, e.g., emotional well-being [ 5 , 6 , 7 , 8 ], sleep tracking [ 9 , 10 ], eating [ 11 ], agitation [ 12 ] and physical activity detection [ 13 , 14 ]. Many works have also focused on identifying behavioral and biometric markers, which can be extracted from such data and provide insights into disciplines such as general medicine [ 15 ] or sports [ 16 ], examining various well-being problems [ 17 , 18 , 19 ].…”

Section: Introductionmentioning

confidence: 99%

E-Prevention: Advanced Support System for Monitoring and Relapse Prevention in Patients with Psychotic Disorders Analyzing Long-Term Multimodal Data from Wearables and Video Captures

Zlatintsi

Filntisis

Garoufis

et al. 2022

Sensors

View full text Add to dashboard Cite

Wearable technologies and digital phenotyping foster unique opportunities for designing novel intelligent electronic services that can address various well-being issues in patients with mental disorders (i.e., schizophrenia and bipolar disorder), thus having the potential to revolutionize psychiatry and its clinical practice. In this paper, we present e-Prevention, an innovative integrated system for medical support that facilitates effective monitoring and relapse prevention in patients with mental disorders. The technologies offered through e-Prevention include: (i) long-term continuous recording of biometric and behavioral indices through a smartwatch; (ii) video recordings of patients while being interviewed by a clinician, using a tablet; (iii) automatic and systematic storage of these data in a dedicated Cloud server and; (iv) the ability of relapse detection and prediction. This paper focuses on the description of the e-Prevention system and the methodologies developed for the identification of feature representations that correlate with and can predict psychopathology and relapses in patients with mental disorders. Specifically, we tackle the problem of relapse detection and prediction using Machine and Deep Learning techniques on all collected data. The results are promising, indicating that such predictions could be made and leading eventually to the prediction of psychopathology and the prevention of relapses.

show abstract

“…Other studies used RNNs for cardiac pathology detection from cardiac signals. For example, B. Ballinger et al predict in [2] cardiovascular risk by detecting one or several pathologies among diabetes, high cholesterol, hypertension and sleep apnea. They developed for this purpose a network model based on 1D-CNN and B-LSTM layers, taking as input different sequences, including the heart rate measured by PPG.…”

Section: Introductionmentioning

confidence: 99%

Gated Recurrent Unit-Based RNN for Remote Photoplethysmography Signal Segmentation

Sabour¹,

Benezeth²

2022

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

View full text Add to dashboard Cite

Remote Photoplethysmography (rPPG) enables quantifying blood volume variations in the skin tissues from an input video recording, using a regular RGB camera. Obtained pulse signals often contain noisy portions due to motion, leading researchers to put aside a great number of rPPG signals in their studies. In this paper, an approach using a Gated Recurrent Unit-based neural network model in order to identify reliable portions in rPPG signals is proposed. This is done by classifying rPPG signal samples into reliable and unreliable samples. For this purpose, rPPG and electrocardiography signals (ECG) were collected from 11 participants, rPPG signal samples were labeled (ECG was used as ground truth), and data were augmented to reach a total number of 11000 1-minute-long rPPG signals. We developed a model composed of a unidimensional CNN and a Bidirectional GRU (1D-CNN+B-GRU) for this study, and obtained an accuracy rate of 85.88%.

show abstract

DeepHeart: Semi-Supervised Sequence Learning for Cardiovascular Risk Prediction

Cited by 83 publications

References 27 publications

High-Resolution Digital Phenotypes From Consumer Wearables and Their Applications in Machine Learning of Cardiometabolic Risk Markers: Cohort Study

High-Resolution Digital Phenotypes From Consumer Wearables and Their Applications in Machine Learning of Cardiometabolic Risk Markers: Cohort Study

E-Prevention: Advanced Support System for Monitoring and Relapse Prevention in Patients with Psychotic Disorders Analyzing Long-Term Multimodal Data from Wearables and Video Captures

Gated Recurrent Unit-Based RNN for Remote Photoplethysmography Signal Segmentation

Contact Info

Product

Resources

About