Publications

Nonverbal vocalizations, such as sighs, grunts, and yells, are informative expressions within typical verbal speech. Likewise, individuals who produce 0-10 spoken words or word approximations ("minimally speaking" individuals) convey rich affective and communicative information through nonverbal vocalizations even without verbal speech. Yet, despite their rich content, little to no data exists on the vocal expressions of this population. Here, we present ReCANVo: Real-World Communicative and Affective Nonverbal Vocalizations - a novel dataset of non-speech vocalizations labeled by function from minimally speaking individuals. The ReCANVo database contains over 7000 vocalizations spanning communicative and affective functions from eight minimally speaking individuals, along with communication profiles for each participant. Vocalizations were recorded in real-world settings and labeled in real-time by a close family member who knew the communicator well and had access to contextual information while labeling. ReCANVo is a novel database of nonverbal vocalizations from minimally speaking individuals, the largest available dataset of nonverbal vocalizations, and one of the only affective speech datasets collected amidst daily life across contexts.

Commercial ultrasound vascular phantoms lack the anatomic diversity required for robust pre-clinical interventional device testing. We fabricated individualized phantoms to test an artificial intelligence enabled ultrasound-guided surgical robotic system (AI-GUIDE) which allows novices to cannulate deep vessels. After segmenting vessels on computed tomography scans, vessel cores, bony anatomy, and a mold tailored to the skin contour were 3D-printed. Vessel cores were coated in silicone, surrounded in tissue-mimicking gel tailored for ultrasound and needle insertion, and dissolved with water. One upper arm and four inguinal phantoms were constructed. Operators used AI-GUIDE to deploy needles into phantom vessels. Two groin phantoms were tested due to imaging artifacts in the other two phantoms. Six operators (medical experience: none, 3; 1-5 years, 2; 5+ years, 1) inserted 27 inguinal needles with 81% (22/27) success in a median of 48 seconds. Seven operators performed 24 arm injections, without tuning the AI for arm anatomy, with 71% (17/24) success. After excluding failures due to motor malfunction and a defective needle, success rate was 100% (22/22) in the groin and 85% (17/20) in the arm. Individualized 3D-printed phantoms permit testing of surgical robotics across a large number of operators and different anatomic sites. AI-GUIDE operators rapidly and reliably inserted a needle into target vessels in the upper arm and groin, even without prior medical training. Virtual device trials in individualized 3-D printed phantoms may improve rigor of results and expedite translation.

We developed vascular phantoms mapped from human subjects to test AI-enabled ultrasound-guided vascular cannulation. Translational device prototyping necessitates anatomically accurate models. Commercial phantoms fail to address anatomic variability. Uniformity leads to optimistic AI model and operator performance. Individualized 3D-printed vascular phantoms yield anatomically correct models optimized for AI-device testing.

Private Automated Contact Tracing (PACT) was a collaborative team and effort formed during the beginning of the Coronavirus Disease 2019 (COVID-19) pandemic. PACT's mission was to enhance contact tracing in pandemic response by designing exposure-detection functions in personal digital communication devices that have maximal public health utility while preserving privacy. This report explains and discusses the use of automated exposure notification during the COVID-19 pandemic and to provide some recommendations for those who may try to design and deploy similar technologies in future pandemics.

In the face of the global pandemic caused by the disease COVID-19, researchers have increasingly turned to simple measures to detect and monitor the presence of the disease in individuals at home. We sought to determine if measures of neuromotor coordination, derived from acoustic time series, as well as phoneme-based and standard acoustic features extracted from recordings of simple speech tasks could aid in detecting the presence of COVID-19. We further hypothesized that these features would aid in characterizing the effect of COVID-19 on speech production systems. A protocol, consisting of a variety of speech tasks, was administered to 12 individuals with COVID-19 and 15 individuals with other viral infections at University Hospital Galway. From these recordings, we extracted a set of acoustic time series representative of speech production subsystems, as well as their univariate statistics. The time series were further utilized to derive correlation-based features, a proxy for speech production motor coordination. We additionally extracted phoneme-based features. These features were used to create machine learning models to distinguish between the COVID-19 positive and other viral infection groups, with respiratory- and laryngeal-based features resulting in the highest performance. Coordination-based features derived from harmonic-to-noise ratio time series from read speech discriminated between the two groups with an area under the ROC curve (AUC) of 0.94. A longitudinal case study of two subjects, one from each group, revealed differences in laryngeal based acoustic features, consistent with observed physiological differences between the two groups. The results from this analysis highlight the promise of using nonintrusive sensing through simple speech recordings for early warning and tracking of COVID-19.

This paper introduces an automated post-traumatic stress disorder (PTSD) screening tool that could potentially be used as a self-assessment or inserted into routine medical visits for PTSD diagnosis and treatment. Methods: With an emotion estimation algorithm providing arousal (excited to calm) and valence (pleasure to displeasure) levels through discourse, we select regions of the acoustic signal that are most salient for PTSD detection. Our algorithm was tested on a subset of data from the DVBIC-TBICoE TBI Study, which contains PTSD Check List Civilian (PCL-C) assessment scores. Results: Speech from low-arousal and positive-valence regions provide the best discrimination for PTSD. Our model achieved an AUC (area under the curve) equal to 0.80 in detecting PCL-C ratings, outperforming models with no emotion filtering (AUC = 0.68). Conclusions: This result suggests that emotion drives the selection of the most salient temporal regions of an audio recording for PTSD detection.

Hemorrhage is the leading cause of preventable death from trauma. Accurate monitoring of hemorrhage and resuscitation can significantly reduce mortality and morbidity but remains a challenge due to the low sensitivity of traditional vital signs in detecting blood loss and possible hemorrhagic shock. Vital signs are not reliable early indicators because of physiological mechanisms that compensate for blood loss and thus do not provide an accurate assessment of volume status. As an alternative, machine learning (ML) algorithms that operate on an arterial blood pressure (ABP) waveform have been shown to provide an effective early indicator. However, these ML approaches lack physiological interpretability. In this paper, we evaluate and compare the performance of ML models trained on nine ABP-derived features that provide physiological insight, using a database of 13 human subjects from a lower-body negative pressure (LBNP) model of progressive central hypovolemia and subsequent progressive restoration to normovolemia (i.e., simulated hemorrhage and whole blood resuscitation). Data were acquired at multiple repressurization rates for each subject to simulate varying resuscitation rates, resulting in 52 total LBNP collections. This work is the first to use a single ABP-based algorithm to monitor both simulated hemorrhage and resuscitation. A gradient-boosted regression tree model trained on only the half-rise to dicrotic notch (HRDN) feature achieved a root-mean-square error (RMSE) of 13%, an R2 of 0.82, and area under the receiver operating characteristic curve of 0.97 for detecting decompensation. This single-feature model's performance compares favorably to previously reported results from more-complex black box machine learning models. This model further provides physiological insight because HRDN represents an approximate measure of the delay between the ABP ejected and reflected wave and therefore is an indication of cardiac and peripheral vascular mechanisms that contribute to the compensatory response to blood loss and replacement.

The COVID-19 pandemic placed unprecedented demands on the global public health systems for disease surveillance and contact tracing. Engineers and scientists recognized that it might be possible to augment the efforts of public health teams, if a system for automated digital contact tracing could be quickly devised and deployed to the population of smartphones. The Private Automated Contact Tracing (PACT) protocol was one of several digital contact tracing proposals offered worldwide. PACT’s mission—to preserve individuals’ privacy and anonymity while enabling them to quickly alert even nearby strangers of a likely risky exposure—was adopted by Google and Apple and realized in the Exposure Notifications (EN) service and API for mobile application development. The Exposure Notifications system, like many digital proximity tools, is based on Bluetooth signal strength estimation, and keeps much of the necessary information and computation on the smartphones themselves. It implemented a decentralized approach to contact tracing: the public health authority, and other governmental authorities, cannot access the records of an individual’s encounters with others; nor is physical location used or shared by the service. Although the service is available on most modern iOS and Android devices, it is not enabled by default; the individual must opt in to use a particular region’s implementation of the service, either by installing the regional app or by enrolling through a menu of regions in the operating system settings. Likewise, individuals must affirm their consent before the service can share anonymized infection status with the regional public health authority, and alert recent close contacts. The widespread availability of Exposure Notifications through Apple and Google’s platforms has made it a de facto world standard. Determining its accuracy and effectiveness as a public health tool has been a subject of intense interest. In July 2020, CDC’s Innovative Technologies Team designated MIT LL and the PACT team as trusted technical advisors on the deployment of private automated contact tracing systems as part of its overall public health response to COVID-19. The Innovative Technologies Team sought to answer the following key question regarding automated contact tracing: Does automated contact tracing have sufficient public health value that it is worthwhile to integrate it at scale into existing and evolving manual contact tracing systems? Rapidly rising caseloads necessitated parallel-path assessment activities of most mature systems at the time. When access to the Google and Apple Exposure Notifications system became available, MIT LL focused the assessment efforts on the systems being built and deployed. There were two immediate and significant challenges to observing and quantifying the performance of the system as a whole: first, the privacy preserving design decisions of PACT and the system implementers denied access to system-level performance metrics, and second, obtaining accurate “ground truth” data about risky encounters in the population, against which to measure the detector performance, would require an unacceptable level of effort and intrusion. Therefore, MIT LL designed a set of parallel research activities to decompose the problem into components that could be assessed quantifiably (Bluetooth sensor performance, algorithm performance, user preferences and behaviors), components that could be assessed qualitatively (potential cybersecurity risks, potential for malicious use), and components that could be modeled based on current and emergent knowledge (population-level effects). The MIT LL research team conducted early assessments of the privacy and security aspects of new EN app implementations and closely reviewed the available system code exercised by the apps, before conducting a series of phone-to-phone data collections both in the laboratory and in simulated real-world conditions. The data from these experiments fed into models and visualization tools created to predict and understand the risk score output of candidate “weights and thresholds” configurations for EN, i.e., to predict the performance of the system as-built against ground truth data for distance and duration of “exposure”. The data and performance predictions from this effort helped to inform the global and local community of practice in making configuration decisions, and can help to predict the performance of future versions of similar tools, or alternative implementations of the current system. We conducted a human factors and usability review of early app user interfaces and messaging from public health, and designed a follow-on large-scale survey to investigate questions about user trust and system adoption decisions. The results of the human factors, user trust, and adoption studies were used by U.S. public health jurisdictions to make adjustments to public-facing communications, and were shared with Apple and Google to improve the user interface. Information gathered from public health experts enabled us to better understand conventional contact tracing workflows and data streams, and we incorporated that information into an agent-based model of “hybrid” contact tracing plus Exposure Notifications. We then combined it with emerging reports on vaccination, mask effectiveness, social interaction, variant transmissibility, and our own data on the sensitivity and specificity of the Bluetooth “dose” estimator, to predict system-level effects under various conditions. Finally, we helped to establish a network of Exposure Notifications “practitioners” in public health, who surfaced desirable system-level key performance indicators (implemented during 2021 and 2022, in the Exposure Notifications Private Analytics system, or ENPA). At the conclusion of the program, many of the initial conditions of the pandemic had changed. The Exposure Notifications service was available to most of the world, but had only been deployed by 28 U.S. states and territories, and had not been adopted by much of the population in those regions. High case rates during the Omicron surge (December 2021 – January 2022) and newly available ENPA data offered the first hints at calculating “real” state-level performance metrics, but those data belong to the states and many are cautious about publishing. Although Google and Apple have stated that Exposure Notifications was designed for COVID-19, and will not be maintained in its current form after the pandemic ends, the public health and engineering communities show clear interest in using the “lessons learned” from Exposure Notifications and other similar solutions to preserve the capabilities developed and prepare better systems for future public health emergencies. The intent of this report is to document the work that has been completed, as well as to inform where the work could be updated or adapted to meet future needs.

Estimations of ankle moments can provide clinically helpful information on the function of lower extremities and further lead to insight on patient rehabilitation and assistive wearable exoskeleton design. Current methods for estimating ankle moments leave room for improvement, with most recent cutting-edge methods relying on machine learning models trained on wearable sEMG and IMU data. While machine learning eliminates many practical challenges that troubled more traditional human body models for this application, we aim to expand on prior work that showed the feasibility of using LSTM models by employing an ensemble of LSTM networks. We present an adaptive weighted LSTM ensemble network and demonstrate its performance during standing, walking, running, and sprinting. Our result show that the LSTM ensemble outperformed every single LSTM model component within the ensemble. Across every activity, the ensemble reduced median root mean squared error (RMSE) by 0.0017-0.0053 N. m/kg, which is 2.7 – 10.3% lower than the best performing single LSTM model. Hypothesis testing revealed that most reductions in RMSE were statistically significant between the ensemble and other single models across all activities and subjects. Future work may analyze different trajectory lengths and different combinations of LSTM submodels within the ensemble.

Individuals who produce few spoken words ("minimally-speaking" individuals) often convey rich affective and communicative information through nonverbal vocalizations, such as grunts, yells, babbles, and monosyllabic expressions. Yet, little data exists on the affective content of the vocal expressions of this population. Here, we present 78,624 arousal and valence ratings of nonverbal vocalizations from the online ReCANVo (Real-World Communicative and Affective Nonverbal Vocalizations) database. This dataset contains over 7,000 vocalizations that have been labeled with their expressive functions (delight, frustration, etc.) from eight minimally-speaking individuals. Our results suggest that raters who have no knowledge of the context or meaning of a nonverbal vocalization are still able to detect arousal and valence differences between different types of vocalizations based on Likert-scale ratings. Moreover, these ratings are consistent with hypothesized arousal and valence rankings for the different vocalization types. Raters are also able to detect arousal and valence differences between different vocalization types within individual speakers. To our knowledge, this is the first large-scale analysis of affective content within nonverbal vocalizations from minimally verbal individuals. These results complement affective computing research of nonverbal vocalizations that occur within typical verbal speech (e.g., grunts, sighs) and serve as a foundation for further understanding of how humans perceive emotions in sounds.