Domain Adaptation Methods for Improving Lab-to-field Generalization of Cocaine Detection using Wearable ECG

Discriminating cocaine use from other sympathomimetics using wearable electrocardiographic (ECG) sensors

BACKGROUND

Our group has established the feasibility of using on-body electrocardiographic (ECG) sensors to detect cocaine use in the human laboratory. The purpose of the current study was to test whether ECG sensors and features are capable of discriminating cocaine use from other non-cocaine sympathomimetics.

METHODS

Eleven subjects with cocaine use disorder wore the Zephyr BioHarness™ 3 chest band under six experimental (drug and non-drug) conditions, including 1) laboratory, intravenous cocaine self-administration, 2) after a single oral dose of methylphenidate, 3) during aerobic exercise, 4) during tobacco use (N=7 who smoked tobacco), and 5) during routine activities of daily inpatient living (unit activity). Three ECG-derived feature sets served as primary outcome measures, including 1) the RR interval (i.e., heart rate), 2) a group of ECG interval proxies (i.e., PR, QS, QT and QTc intervals), and 3) the full ECG waveform. Discriminatory power between cocaine and non-cocaine conditions for each of the three outcomes measures was expressed as the area under the receiver operating characteristics (AUROC) curve.

RESULTS

All three outcomes successfully discriminated cocaine use from unit activity, exercise, tobacco, and methylphenidate conditions with a mean AUROC values ranging from 0.66 to 0.99 and with least squares means values all statistically different/higher than 0.5 among all subjects [F(3, 99) = 3.38, p =0.02] and among those with tobacco use [F(4, 84) = 5.39, p = 0.0007].

CONCLUSIONS

These preliminary results support discriminatory power of wearable ECG sensors for detecting cocaine use.

Domain Adaptation Methods for Lab-to-Field Human Context Recognition

Human context recognition (HCR) using sensor data is a crucial task in Context-Aware (CA) applications in domains such as healthcare and security. Supervised machine learning HCR models are trained using smartphone HCR datasets that are scripted or gathered in-the-wild. Scripted datasets are most accurate because of their consistent visit patterns. Supervised machine learning HCR models perform well on scripted datasets but poorly on realistic data. In-the-wild datasets are more realistic, but cause HCR models to perform worse due to data imbalance, missing or incorrect labels, and a wide variety of phone placements and device types. Lab-to-field approaches learn a robust data representation from a scripted, high-fidelity dataset, which is then used for enhancing performance on a noisy, in-the-wild dataset with similar labels. This research introduces Triplet-based Domain Adaptation for Context REcognition (Triple-DARE), a lab-to-field neural network method that combines three unique loss functions to enhance intra-class compactness and inter-class separation within the embedding space of multi-labeled datasets: (1) domain alignment loss in order to learn domain-invariant embeddings; (2) classification loss to preserve task-discriminative features; and (3) joint fusion triplet loss. Rigorous evaluations showed that Triple-DARE achieved 6.3% and 4.5% higher F1-score and classification, respectively, than state-of-the-art HCR baselines and outperformed non-adaptive HCR models by 44.6% and 10.7%, respectively.

Systematic review: Wearable remote monitoring to detect nonalcohol/nonnicotine-related substance use disorder symptoms

BACKGROUND AND OBJECTIVES

Substance use disorders (SUDs) are chronic relapsing diseases characterized by significant morbidity and mortality. Phenomenologically, patients with SUDs present with a repeating cycle of intoxication, withdrawal, and craving, significantly impacting their diagnosis and treatment. There is a need for better identification and monitoring of these disease states. Remote monitoring chronic illness with wearable devices offers a passive, unobtrusive, constant physiological data assessment. We evaluate the current evidence base for remote monitoring of nonalcohol, nonnicotine SUDs.

METHODS

We performed a systematic, comprehensive literature review and screened 1942 papers.

RESULTS

We found 15 studies that focused mainly on the intoxication stage of SUD. These studies used wearable sensors measuring several physiological parameters (ECG, HR, O2 , Accelerometer, EDA, temperature) and implemented study-specific algorithms to evaluate the data.

DISCUSSION AND CONCLUSIONS

Studies were extracted, organized, and analyzed based on the three SUD disease states. The sample sizes were relatively small, focused primarily on the intoxication stage, had low monitoring compliance, and required significant computational power preventing "real-time" results. Cardiovascular data was the most consistently valuable data in the predictive algorithms. This review demonstrates that there is currently insufficient evidence to support remote monitoring of SUDs through wearable devices.

SCIENTIFIC SIGNIFICANCE

This is the first systematic review to show the available data on wearable remote monitoring of SUD symptoms in each stage of the disease cycle. This clinically relevant approach demonstrates what we know and do not know about the remote monitoring of SUDs within disease states.

OpiTrack: A Wearable-based Clinical Opioid Use Tracker with Temporal Convolutional Attention Networks

Opioid use disorder is a medical condition with major social and economic consequences. While ubiquitous physiological sensing technologies have been widely adopted and extensively used to monitor day-to-day activities and deliver targeted interventions to improve human health, the use of these technologies to detect drug use in natural environments has been largely underexplored. The long-term goal of our work is to develop a mobile technology system that can identify high-risk opioid-related events (i.e., development of tolerance in the setting of prescription opioid use, return-to-use events in the setting of opioid use disorder) and deploy just-in-time interventions to mitigate the risk of overdose morbidity and mortality. In the current paper, we take an initial step by asking a crucial question: Can opioid use be detected using physiological signals obtained from a wrist-mounted sensor? Thirty-six individuals who were admitted to the hospital for an acute painful condition and received opioid analgesics as part of their clinical care were enrolled. Subjects wore a noninvasive wrist sensor during this time (1-14 days) that continuously measured physiological signals (heart rate, skin temperature, accelerometry, electrodermal activity, and interbeat interval). We collected a total of 2070 hours (≈ 86 days) of physiological data and observed a total of 339 opioid administrations. Our results are encouraging and show that using a Channel-Temporal Attention TCN (CTA-TCN) model, we can detect an opioid administration in a time-window with an F1-score of 0.80, a specificity of 0.77, sensitivity of 0.80, and an AUC of 0.77. We also predict the exact moment of administration in this time-window with a normalized mean absolute error of 8.6% and R ² coefficient of 0.85.

Automatic Detection of Atrial Fibrillation in ECG Using Co-Occurrence Patterns of Dynamic Symbol Assignment and Machine Learning

Early detection of atrial fibrillation from electrocardiography (ECG) plays a vital role in the timely prevention and diagnosis of cardiovascular diseases. Various algorithms have been proposed; however, they are lacking in considering varied-length signals, morphological transitions, and abnormalities over long-term recordings. We propose dynamic symbolic assignment (DSA) to differentiate a normal sinus rhythm (SR) from paroxysmal atrial fibrillation (PAF). We use ECG signals and their interbeat (RR) intervals from two public databases namely, AF Prediction Challenge Database (AFPDB) and AF Termination Challenge Database (AFTDB). We transform RR intervals into a symbolic representation and compute co-occurrence matrices. The DSA feature is extracted using varied symbol-length V, word-size W, and applied to five machine learning algorithms for classification. We test five hypotheses: (i) DSA captures the dynamics of the series, (ii) DSA is a reliable technique for various databases, (iii) optimal parameters improve DSA’s performance, (iv) DSA is consistent for variable signal lengths, and (v) DSA supports cross-data analysis. Our method captures the transition patterns of the RR intervals. The DSA feature exhibit a statistically significant difference in SR and PAF conditions (p < 0.005). The DSA feature with W=3 and V=3 yield maximum performance. In terms of F-measure (F), rotation forest and ensemble learning classifier are the most accurate for AFPDB (F = 94.6%) and AFTDB (F = 99.8%). Our method is effective for short-length signals and supports cross-data analysis. The DSA is capable of capturing the dynamics of varied-lengths ECG signals. Particularly, the optimal parameters-based DSA feature and ensemble learning could help to detect PAF in long-term ECG signals. Our method maps time series into a symbolic representation and identifies abnormalities in noisy, varied-length, and pathological ECG signals.

An Examination of the Feasibility of Detecting Cocaine Use Using Smartwatches

As digital technology increasingly informs clinical trials, novel ways to collect study data in the natural field setting have the potential to enhance the richness of research data. Cocaine use in clinical trials is usually collected via self-report and/or urine drug screen results, both of which have limitations. This article examines the feasibility of developing a wrist-worn device that can detect sufficient physiological data (i.e., heart rate and heart rate variability) to detect cocaine use. This study aimed to develop a wrist-worn device that can be used in the natural field setting among people who use cocaine to collect reliable data (determined by data yield, device wearability, and data quality) that is less obtrusive than chest-based devices used in prior research. The study also aimed to further develop a cocaine use detection algorithm used in previous research with an electrocardiogram on a chestband by adapting it to a photoplethysmography sensor on the wrist-worn device which is more prone to motion artifacts. Results indicate that wrist-based heart rate data collection is feasible and can provide higher data yield than chest-based sensors, as wrist-based devices were also more comfortable and affected participants' daily lives less often than chest-based sensors. When properly worn, wrist-based sensors produced similar quality of heart rate and heart rate variability features to chest-based sensors and matched their performance in automated detection of cocaine use events. Clinical Trial Registration: www.ClinicalTrials.gov, identifier: NCT02915341.

Unsupervised Domain Adaptation for ECG Arrhythmia Classification

Electrocardiograph (ECG) is one of the most critical physiological signals for arrhythmia diagnosis in clinical practice. In recent years, various algorithms based on deep learning have been proposed to solve the heartbeat classification problem and achieved saturated accuracy in intrapatient paradigm, but encountered performance degradation in inter-patient paradigm due to the drastic variation of ECG signals among different individuals. In this paper, we propose a novel unsupervised domain adaptation scheme to address this problem. Specifically, we first propose a robust baseline model called Multi-path Atrous Convolutional Network (MACN) to tackle ECG heartbeat classification. Further, we introduce Cluster-aligning loss and Cluster-separating loss to align the distributions of training and test data and increase the discriminability, respectively. The proposed method requires no expert annotations but a short period of unlabelled data in new records. Experimental results on the MIT-BIH database demonstrate that our scheme effectively intensifies the baseline model and achieves competitive performance with other state-of-the-arts.

Combining ecological momentary assessment with objective, ambulatory measures of behavior and physiology in substance-use research

Whereas substance-use researchers have long combined self-report with objective measures of behavior and physiology inside the laboratory, developments in mobile/wearable electronic technology are increasingly allowing for the collection of both subjective and objective information in participants' daily lives. For self-report, ecological momentary assessment (EMA), as implemented on contemporary smartphones or personal digital assistants, can provide researchers with near-real-time information on participants' behavior and mood in their natural environments. Data from portable/wearable electronic sensors measuring participants' internal and external environments can be combined with EMA (e.g., by timestamps recorded on questionnaires) to provide objective information useful in determining the momentary context of behavior and mood and/or validating participants' self-reports. Here, we review three objective ambulatory monitoring techniques that have been combined with EMA, with a focus on detecting drug use and/or measuring the behavioral or physiological correlates of mental events (i.e., emotions, cognitions): (1) collection and processing of biological samples in the field to measure drug use or participants' physiological activity (e.g., hypothalamic-pituitary-adrenal axis activity); (2) global positioning system (GPS) location information to link environmental characteristics (disorder/disadvantage, retail drug outlets) to drug use and affect; (3) ambulatory electronic physiological monitoring (e.g., electrocardiography) to detect drug use and mental events, as advances in machine learning algorithms make it possible to distinguish target changes from confounds (e.g., physical activity). Finally, we consider several other mobile/wearable technologies that hold promise to be combined with EMA, as well as potential challenges faced by researchers working with multiple mobile/wearable technologies simultaneously in the field.

Automatic Detection of Opioid Intake Using Wearable Biosensor

A plethora of research shows that recreational drug overdoses result in major social and economic consequences. However, current illicit drug use detection in forensic toxicology is delayed and potentially compromised due to lengthy sample preparation and its subjective nature. With this in mind, scientists have been searching for ways to create a fast and easy method to detect recreational drug use. Therefore, we have developed a method for automatic detection of opioid intake using electrodermal activity (EDA), skin temperature and tri-axis acceleration data generated from a wrist worn biosensor. The proposed system can be used for home and hospital use. We performed supervised learning and extracted 23 features using time and frequency domain analysis to recognize pre- and post- opioid health conditions in patients. Feature selection procedures are used to reduce the number of features and processing time. For supervised learning, we compared three classifiers and selected the one with highest accuracy and sensitivity: decision tree, k-nearest neighbors (KNN) and eXtreme Gradient Boosting utilizing modified features. The results show that the proposed method can detect opioid use in real-time with 99% accuracy. Moreover, this method can be applied to identify other use of additional substances other than opioids. The numerical analysis is completed on data collected from 30 participants over a span of 4 months.

rConverse: Moment by Moment Conversation Detection Using a Mobile Respiration Sensor

Monitoring of in-person conversations has largely been done using acoustic sensors. In this paper, we propose a new method to detect moment-by-moment conversation episodes by analyzing breathing patterns captured by a mobile respiration sensor. Since breathing is affected by physical and cognitive activities, we develop a comprehensive method for cleaning, screening, and analyzing noisy respiration data captured in the field environment at individual breath cycle level. Using training data collected from a speech dynamics lab study with 12 participants, we show that our algorithm can identify each respiration cycle with 96.34% accuracy even in presence of walking. We present a Conditional Random Field, Context-Free Grammar (CRF-CFG) based conversation model, called rConverse, to classify respiration cycles into speech or non-speech, and subsequently infer conversation episodes. Our model achieves 82.7% accuracy for speech/non-speech classification and it identifies conversation episodes with 95.9% accuracy on lab data using a leave-one-subject-out cross-validation. Finally, the system is validated against audio ground-truth in a field study with 32 participants. rConverse identifies conversation episodes with 71.7% accuracy on 254 hours of field data. For comparison, the accuracy from a high-quality audio-recorder on the same data is 71.9%.

mHealth for the Detection and Intervention in Adolescent and Young Adult Substance Use Disorder

Purpose of review

The goal of this review is to highlight recent research in mHealth based approaches to the detection and treatment of substance use disorders in adolescents and young adults.

Recent findings

The main methods for mHealth based detection include mobile phone based self-report tools, GPS tracking, and wearable sensors. Wearables can be used to detect physiologic changes (e.g., heart rate, electrodermal activity) or biochemical contents of analytes (i.e. alcohol in sweat) with reasonable accuracy, but larger studies are needed. Detection methods have been combined with interventions based on mindfulness, education, incentives/goals and motivation. Few studies have focused specifically on the young adult population, although those that did indicate high rates of utilization and acceptance.

Summary

Research that explores the pairing of advanced detection methods such as wearables with real time intervention strategies is crucial to realizing the full potential of mHealth in this population.

Wearable ballistocardiogram and seismocardiogram systems for health and performance

Cardiovascular diseases (CVDs) are prevalent in the US, and many forms of CVD primarily affect the mechanical aspects of heart function. Wearable technologies for monitoring the mechanical health of the heart and vasculature could enable proactive management of CVDs through titration of care based on physiological status as well as preventative wellness monitoring to help promote lifestyle choices that reduce the overall risk of developing CVDs. Additionally, such wearable technologies could be used to optimize human performance in austere environments. This review describes our progress in developing wearable ballistocardiogram (BCG)- and seismocardiogram-based systems for monitoring relative changes in cardiac output, contractility, and blood pressure. Our systems use miniature, low-noise accelerometers to measure the movements of the body in response to the heartbeat and novel machine learning algorithms to provide robustness against motion artifacts and sensor misplacement. Moreover, we have mathematically related wearable BCG signals-representing local, cardiogenic movements of a point on the body-to better understood whole body BCG signals, and thereby improved estimation of key health parameters. We validated these systems with experiments in healthy subjects, studies in patients with heart failure, and measurements in austere environments such as water immersion. The systems can be used in future work as a tool for clinicians and physiologists to measure the mechanical aspects of cardiovascular function outside of clinical settings, and to thereby titrate care for patients with CVDs, provide preventative screening, and optimize performance in austere environments by providing real-time in-depth information regarding performance and risk.