coe-staff: Save the Date: SDS Network Invited Colloquium

[Lisa Fortin]

[https://files.constantcontact.com/e2afc7af701/269d0cec-05d8-4c9e-9c04-a61acc084b7f.png]


Save The Date

SDS Network<https://blogs.uoregon.edu/sdsnetwork/> Invited Colloquium
Deep Learning for Predicting Depression and
Anxiety from Spoken Language
Dr. Elizabeth Shriberg
October 16, 2020
1:30-3:00 p.m. PST

RSVP<https://urldefense.com/v3/__https:/oregon.qualtrics.com/jfe/form/SV_bx4862NYcmcnNg9__;!!C5qS4YX3!XOuDAPFLxffMTm3Bs5-MN0EESFaOjwxJj5DQioA89VGClJ10O5RhT0dB6ihwsd4$>




Abstract: Depression and anxiety are globally prevalent and debilitating conditions that are often under-diagnosed. Digital health solutions can play a critical role in the management of behavioral health by scaling capacity for screening and for remote monitoring of patients between provider visits. Speech technology offers promise because speaking is natural, engaging, carries meaning, and requires only a microphone.

This talk describes recent work at Ellipsis Health, a San Francisco start-up, on two deep learning models developed for this purpose. One model uses natural language processing. The other uses speech acoustics. Both models employ transfer learning. Speech data come from two corpora representing a total of over 11,000 users who interacted with an application. PHQ-8 and GAD-7 labels are used for supervised training and model evaluation.

Results are presented for both depression and anxiety, showing that binary classification rates are near or above 0.80 AUC. Further analyses explore the effect of training data size, effect of test sample length, and portability over different patient demographics such as age, gender, and ethnicity. Finally, the talk shares some lessons learned and remaining challenges for real-word deployments in this domain.



[https://files.constantcontact.com/e2afc7af701/f366e75a-933d-41f5-8007-b40caddd5ea3.jpg]

Dr. Elizabeth Shriberg is a speech scientist with over 25 years of experience in the analysis and computational modeling of spoken language. She is currently Chief Science Officer at Ellipsis Health, a San Francisco start-up developing artificial intelligence solutions based on spoken conversation analysis, to advance behavioral health. She is also an Affiliate of Johns Hopkins University / HLTCOE in Baltimore, MD, and an External Fellow of the International Computer Science Institute in Berkeley, CA. Prior to Ellipsis she was a Principal Scientist at SRI International in Menlo Park, CA, leading data science and machine learning teams working on government and commercial R&D efforts in speech-based emotion and affective computing, health monitoring, speech understanding, dialog modeling, computational prosody, speaker verification, and modeling of speech disfluencies. Dr. Shriberg was also a Principal Researcher at Microsoft Research in Mountain View, CA, developing methods for natural conversation with machines. She has published over 300 papers and patents and is a Fellow of the International Speech Communication Association and of SRI International. Dr. Shriberg serves on international academic association, conference, and journal boards. She also advises multiple startups working in conversational AI.





[Facebook]‌ <https://urldefense.com/v3/__https:/www.facebook.com/uoeducation__;!!C5qS4YX3!XOuDAPFLxffMTm3Bs5-MN0EESFaOjwxJj5DQioA89VGClJ10O5RhT0dBtLy9ltk$> [Twitter] ‌ <https://urldefense.com/v3/__https:/twitter.com/uoeducation__;!!C5qS4YX3!XOuDAPFLxffMTm3Bs5-MN0EESFaOjwxJj5DQioA89VGClJ10O5RhT0dByDYkbTg$> [Instagram] ‌ <https://urldefense.com/v3/__https:/www.instagram.com/uoeducation/__;!!C5qS4YX3!XOuDAPFLxffMTm3Bs5-MN0EESFaOjwxJj5DQioA89VGClJ10O5RhT0dBofgpE_E$> [LinkedIn] ‌ <https://urldefense.com/v3/__https:/www.linkedin.com/company/uo-college-of-education__;!!C5qS4YX3!XOuDAPFLxffMTm3Bs5-MN0EESFaOjwxJj5DQioA89VGClJ10O5RhT0dBC5DxW7g$> [YouTube] ‌ <https://urldefense.com/v3/__https:/www.youtube.com/user/uoeducation__;!!C5qS4YX3!XOuDAPFLxffMTm3Bs5-MN0EESFaOjwxJj5DQioA89VGClJ10O5RhT0dBNRt3l34$>


University of Oregon College of Education | 1655 Alder Street, Eugene, OR 97403




-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.uoregon.edu/pipermail/coe-staff/attachments/20201002/28d1ca7c/attachment.html>