We propose to build open source hardware and open source software to collect and analyze voice data to help diagnose and assess the severity of selective mutism, a disorder where affected individuals are mute in specific contexts but are able to speak in other specific contexts, and possibly of other anxiety disorders. The Child Mind Institute has one of the largest selective mutism clinics in the world, and we have experience recording voice data from children with selective mutism in a laboratory setting, in the clinic, as well as at home via a wearable voice recorder. We will design new hardware, software, and experimental designs that will record voice and other data to better understand what conditions impede affected individuals’ ability to speak.
For data collection, we will build a new mobile voice recorder. For data analysis, we will build on a voice feature extraction and analysis software pipeline that we initially developed for use in tracking symptom severity of Parkinson’s disease. This research promises to empower people with anxiety disorders such as selective mutism by providing feedback about their condition, with the ultimate goal of using patterns of voice features and other contextualizing data to diagnose, monitor, and predict behaviors associated with anxiety. Users, parents, and care providers will all be able to use their own data and feedback with appropriate privacy settings.
Based on our ongoing analysis of voice recordings collected from study participants with and without selective mutism, measurable patterns in the sound recordings appear to be indicative of the disorder. These findings align with the research of Evelyn Klein and Cesar Ruiz at LaSalle University, whose findings indicate measurable differences in throat electromyography between individuals with and without selective mutism (Chesney, 2015/2016; Klein & Ruiz, 2017). Additionally, research by Alex Cohen at Louisiana State University, among others, shows promise for machine learning techniques for objectively detecting symptoms of mental health disorders from voice data (Cohen & Elvevåg, 2014; Cohen, Mitchell, & Elvevåg, 2014; Cohen, Renshaw, Mitchell, & Kim, 2016).
In prior voice analysis research, we have used mobile sensors, feature extraction, and prediction models to enable a scalable approach for estimating individual variation in depression and Parkinson’s disease (Ghosh, 2016) and helped to develop the mPower smartphone app on top of Apple’s ResearchKit (Bot et al., 2016). The latter included voice analysis in its tracking of cognitive, behavioral, mood, and physiological states in Parkinson’s disease (Chaibub Neto et al., 2017). We have used an ecological audio recorder “to track and quantify progress of children enrolled in Brave Buddies, a week-long intensive group behavioral treatment program designed by the Child Mind Institute” (Busman et al., 2016). In addition to our voice recording of patients with selective mutism, we are also collecting audio, video, and biometric recordings as part of the Child Mind Institute’s Healthy Brain Network study of 10,000 children.
Early research in selective mutism at the Child Mind Institute indicate the presence of localized somatic symptoms of the disorder. Times in which persons with selective mutism attempt to speak but are unable to do so are of particular interest. The proposed research will advance understanding of both the mechanisms and phenotypic indicators of selective mutism, as well as improving the ability to accurately diagnose and track the disorder. Our open source, wearable audio recording device and open source voice analysis software pipeline for general feature extraction and analysis of voice data could be used for scientific research and user self-monitoring for any mental health condition.
We recently collected data from a selective mutism population and a population of control participants using both manual coding and a LENA Digital Language Processor in a Parent–Child Interaction Therapy-based experimental setting. Researchers here at the Child Mind Institute and elsewhere have been working to develop methods of combining acoustic, contextual and content-focused voice data to algorithmically measure symptoms of psychological disorders. Moving forward, we would like to combine the best strengths and resources of each study and institution. At the Child Mind Institute, we in the Center for the Developing Brain have appropriate facilities, knowledge, experience and access to the Selective Mutism Service’s world-class team and unrivaled population of our target population.
For the proposed research, we would upgrade our data collection devices to include a microphone array (to replace the LENA Digital Language Processor) and a device that measures either electrodermal activity (to infer stress level) or throat movements (to replace the tethered desktop electromyographs used at LaSalle University for detecting swallowing behavior) in a Parent–Child Interaction Therapy-based experimental setting. In our revised experimental design, we would collect longitudinal data from as many participants as possible in order to evaluate our model’s ability to measure changes in symptoms over time.
We will collect a rich set of data, including high-, mid- and low-level voice (e.g., speech content, speech cadence and fundamental frequency, respectively) and contextual audio data, electrodermal or swallowing data, researcher observations and participant diagnoses. We will isolate low-level features relevant to diagnosing symptoms of selective mutism and will also begin exploring voice data for correlates to selective mutism diagnoses in mid-level and high-level voice data and in contextual data.
Because of the contextual and temporal qualities of selective mutism’s diagnostic features, discovery of quick and reliable indicators of the physical and behavioral symptoms of the disorder will have profound significance in both the diagnosis of selective mutism and in tracking progress of people recovering from the disorder. In current practice, the disorder is likely underdetected due to its contextual nature and thereby underdiagnosed; its negative effects can be subtle but compounding and long-lasting. By determining measurable physical evidence of the disorder’s presence, selective mutism can be addressed more rapidly both in individual cases and in medical understanding. An eventual use-case would involve persistent ecological tracking for users of the device produced by this project, something akin to a Fitbit targeted at our diagnostic features. The voice analysis resources we will create and employ will also benefit clinical research in other mental health disorders as well.
According to http://selectivemutism.org/, DSM-IV-TR estimates that SM affects 1 in 1000 children referred for mental health treatment (APA, 2000). However, several researchers have suggested that the true prevalence of SM in the general population is largely underestimated (Bergman et al., 2002; Hayden, 1980; Hesselman, 1983; Kupietz & Schwartz, 1982; & Thompson, 1988). Recent studies show that SM is not as rare as it was previously believed to be but is comparable to other, widely known disorders of childhood. A study targeting a large sample of children in a Los Angeles, CA school district identified children who met the diagnostic criteria for SM and found a prevalence rate of 7.1 per 1,000 children (Bergman et al., 2002). A subsequent study in Israel found an almost identical prevalence rate (Elizur & Perednik, 2003). These numbers suggest that SM has a higher prevalence than autism (.5 per 1000), major depressive disorder (.4 to 3 per 1000), Tourette’s disorder (.5 per 1000), obsessive-compulsive disorder (.5 to 1 in 1000) and other well-known disorders. In comparison to other studies, which only accounted for diagnosed cases of SM, provides evidence that a large number of individuals with SM are undiagnosed or misdiagnosed. Parents of children with SM who enter treatment often report that their child was misdiagnosed with autism or another pervasive developmental disorder, mental retardation or oppositional-defiant disorder. Most are told (if anything) by uniformed professionals that there is nothing wrong with their child, that their child is ‘just shy,’ or will grow out of this behavior. Thus, the lack of awareness among educators and treating professionals leads to delays in diagnosis and missed opportunities for treatment.
More recent estimates of prevalence place “point prevalence . . . between 0.03% and 1% depending on the setting (e.g., clinic vs. school vs. general population) and ages of the individuals in the sample. The prevalence of the disorder does not seem to vary by sex or race / ethnicity. The disorder is more likely to manifest in young children than in adolescents and adults” (APA, 2013, p. 196). The American Speech-Language-Hearing Association describes the current difficulty in accurately calculating the prevalence of the disorder, stating on asha.org, “Accurate population estimates of selective mutism have been difficult to ascertain due to the relative rarity of the condition, differences in sampled populations, variations in diagnostic procedures (e.g., chart review, standardized assessment), and the use of different diagnostic criteria (Busse & Downey, 2011; Sharkey & McNicholas, 2008; Viana, Beidel, & Rabian, 2009).”