Project Psi-Fi (??) leverages the Real-time, Open, Portable, Extensible, Speech Lab (OSP) developed under grant R01DC015436 and expands its capabilities as a high fidelity sensor for physical and social Isolation and loneliness assessment in support of clinical investigations focused on Alzheimer?s disease and related dementias (AD/ADRD). There is emerging evidence that sensory (e.g., hearing loss), motor, and/or cognitive impairments may contribute to limited social interactions, reduced physical activity, and ensuing perception of loneliness. Project ?? proposes to build software tools for quantifying oral interactions and physical activities during activities of daily living (ADL) and augment these objective metrics with subjective assessment of perceived loneliness using ecological momentary assessments (EMA). OSP hardware includes a pager-sized processing and communication device (PCD) with two onboard microphones and a six-degrees-of-freedom inertial motion sensing unit (IMU). OSP software includes real-time signal processing libraries for master hearing aid functions and an embedded web server hosting web apps for users and researchers to configure the OSP and access data from any browser-enabled device. The proposed work extends OSP software as a ?? sensor and enables research in the interplay between physical activity, social isolation, and perceived loneliness during ADL. This project will extend work on advanced signal processing functions done under the parent R01 grant to develop libraries and application programmer interfaces (APIs) for (i) speaker diarization and (ii) physical activity monitoring. Speaker diarization is an emerging discipline in speech processing to determine ?who spoke when? during unstructured oral interactions with unknown number of participants. Project Psi-Fi will adopt speaker diarization algorithms to quantify social interactions based on speech during ADL. Many mature technologies exist to quantify physical activity and sedentary behavior using IMU sensors during ADL. The existing OSP software will be extended to allow physical activity monitoring with geolocations overlay. Existing web apps will also be adapted to administer EMA assessments such as the UCLA loneliness scale for assessing perceived loneliness. APIs will be provided to ?? sensor data and enable researchers to tune the frequency and timing of EMA surveys dynamically, based on the observed users? activity. This project will also extend the work on validation of OSP features for clinical investigations that was done as part of the parent R01 to establish feasibility of a ?? sensor to investigate interactions between physical activity, social engagement, and perceived loneliness among normal and hearing loss subjects.
Project Psi-fi expands the applications of open-source speech platform (OSP) for objective measurements of physical activity, and social interactions during activities of daily living; along with assessment of perceived loneliness using ecological momentary assessments. This project has applications in the realm of hearing loss, Alzheimer's research, and related cognitive disorders. OSP was developed under NIH grant R01DC015436 to enable hearing healthcare researchers with open source tools to advance hearing aids research and to enable psychophysical investigations beyond what is possible today.
| Lee, Ching-Hua; Kates, James M; Rao, Bhaskar D et al. (2017) Speech quality and stable gain trade-offs in adaptive feedback cancellation for hearing aids. J Acoust Soc Am 142:EL388 |
