Pei-yun S. (Sabrina) Hsueh  Pei-yun S. (Sabrina) Hsueh photo       

contact information

Behavioral Analytics, Cognitive Learning and Adaptation, Multimdoal and Text analytics, Human Computer Interaction
Thomas J. Watson Research Center, Yorktown Heights, NY USA
  +1dash914dash945dash1916

links


profile


The overarching theme of my informatics interest ties closely to the marriage of science of care and science of data, with a focus on bridging the gap between population-level evidence and individual patient need. The recent rise of consumer health awareness and the increasing availability of pervasive health technologies have offered new promise to advance precision health research, to better inform patient communication and shared decision making, and to improve patient activation and engagement. The key is to transform the accumulated patient-generated health data (i.e., PGHD) and N-of-1 personal health data (i.e., “small data”) into actionable insights that can enable dynamic evidence delivery to meet patients where they are.

However, the vision has incurred multi-level challenges.

First, the patients’ care team members need to be able to manage rapidly increasing amounts of patient-generated data. Therefore, consumer health information management and analytics tools are needed to filter noisy and irrelevant data, identify insights from multi-layer query and present the minimal description for the care team based on patients’ need in different contexts.

Second, new patient engagement and behavioral change platforms and tools are needed to facilitate patients themselves to understand their own health/wellness status, disease inflection points and progression patterns.

Last but not least, new policy initiatives and data/research network, as an incentive and safeguard mechanism, are needed to increase the interoperability of consumer health data. 

My research thus lies on the intersection of these challenges and the development of a cognitive learning framework that can leverage PGHD and self-experimentation to induce patient engagement and intervention adaptation strategies. First, how to assess applicability and adapt population-level evidence to an individual level? Second, with all the added nuances and patient preferences uncovered in the sequential observations of PGHD, can we generate feedback for users to make sense of their data, or capture individual predictive pathways that can help infer patient need for their care team?   Finally, by combining multiple N-of-1 sequential observations, how to perform meta-analytics and infer best practice strategies as the basis of collective real-world evidence? At the individual level, how to leverage such insights on best practice to further break down an overarching goal into a series of smaller actionable tasks for individuals to implement in intermediate stages?   

To fully capture the benefits of interpreting patient need, these are the questions we need to address. In some cases where we don’t have enough personal health data to infer patient status and preferences, we need to establish an active self-learning mechanism to interactively guide users through a self-experimentation process similar to what N-of-1 trials are attempting to achieve. The ultimate goal is to optimize the tailoring of care plans based on the ideographic understanding of patients so as to bridge the gap between population-level evidence and best practice in real-world evidence from PGHD in an interactive manner. Additional benefits include enabling healthcare professionals and care team to think “outside-of-the box” when trying to find an alternative solution to combat treatment adherence issues.

 

This investigation can also enable us to start collecting evidence on best practice strategies for further investigation. One direction is to produce innovative approaches of computing personalization and incorporating personalization analytics into service design. In the healthcare domain, I am working on active characterization of personal wellness status and active recommendation that are driven by outcome prediction. The importance of personalization research and system design arises from the need of serving the long tail of user need. While many multi-year structured programs have verified the effectiveness of individualized intervention on preventive care and chronic disease management (Helmrich et al, 1991; Bailey, 2001; Finland National Type II Diabetes Prevention Programme, 2007; CDC Diabetes Prevention Program, 2008), the task of offering personalized services dynamically in users’ context has posed grand challenges to existing service providers. On the one hand, integrated care models have shown promises in satisfying the long tail of demand. On the other hand, the success of an integrated model in clinical trials is not enough to secure a disruption in the service market. In fact, its reliance on the constant updates of user wellness status and tailoring of intervention accordingly requires solid system support from both the vendors and system operators. Some key competencies to be provided include: (1) inferring risks from multiple heterogeneous sources (and whenever necessary, going back to the user and care team to solicit for more information); (2) handling multi-faceted risk stratification; and (3) “on the fly” assessment and recommendation with respect to trends shown in the incoming data stream. During the process of data analytics, I am also quite interested in using social signals to improve compliance feedback strategy.

I am involved in the development of an evidence-based wellness management platform in a cloud computing environment. The platform provides an API for healthcare applications to (i) integrate information from heterogeneous data source (Sense), (ii) draw predictions by applying or extending models in a repository (Predict), and (iii) trigger proper responses (Respond). The development side of goal is to enable any independent software vendor (ISV) to use the API and the Sense-Predict-Respond framework to implement their services and exchange information with 3rd party applications.

My roles in IBM Watson Research Center include: 

- IBM Academy of Technology Member 2017-present

- IBM Health behavioral insights lead 2016-present

- IBM World Wide Research Health Informatics Professional Interest Community 2015-present

- Global GTO Healthcare topic industries Co-Lead 2014

- Mobile-First Far Reaching Research Tech Lead 2014

- Wellness analytics Lead 2013-present

- Research Staff Member in Computational Behavior and Decision Science Group

- Research Scientist in Healthcare Transformation

 

My current projects include: 

 - Health Behavioral Insights

> Understand behavioral profiling and explore interpretable cognitive learning methods

> Cognitive care management

 - Personalized system of Insights

 > Design hypothesis-driven exogenous data analytics framework for enterprise data curation, consumption and cross-layer clinical/consumer insight generation

 > Execution of adherence behavior adherence modeling and the design of prospective study in the context of PHM for self-ensured employers

> Liaison with Mobility Competency Center on iOS wellness app development and develop healthcare use cases with wearables and biosensors.

> Ecosystem building and Client/partner relationship management

> Technology consultation for outcome-based business models with partners/clients

> Identify strategies to increase ecosystem value through technology initiatives, assess

technical feasibility and strategic options enabled by new technologies

> IP portfolio/Thought leadership (liaison to Science & Technology department)

> Personalized healthcare platform and mobile applications

> Wearable/IOT/bio-sensor application in healthcare/wellness

> “Precision medicine at Nano-scale”

 

My past projects include: 

- Analytics Lead of In-market Experiment, Taiwan Collaboratory

> Design personalization analytics on Wellness Cloud

> Enable personalized services with clinical insight generation, sampling, context-aware recommendation, adherence monitoring and adaptation.

> Develop AaaS (Analytics-as-a-Service) to deploy insights to 3rdparty SP

> Lead the development of health literacy tool/app (dynamic accretion of patient engagement instruments with collaborative crowdsourcing)

 

Social media analytics (Trend detection from crowd-sourcing data)

Predictive Modeling Group, Business Analytics and Math Science Department

> Social Media Analytics for marketing intelligence

> Blog Analysis of Network Topology and Evolving Responses (BANTER)

> Mining crowd wisdom from unstructured data sources (w. Amazon Mechanical Turk)

> Patent Quality Index for legal communication

> Statistical analysis of quality-indicative features in patents applications/transactions

> Big-data analytics (sampling for natural language processing)

 

Overarching Theme & Previous Work

 

The overarching them of my research interest ties closely to the marriage of artificial intelligence and human computer interaction, with a focus on integrating machine learning and empirical analysis approaches for natural language understanding. My previous research concerns the development of spoken language understanding applications in spontaneous speech, using a variety of approaches ranging from statistical analysis, empirical study to machine learning. This is no secret that people speak differently under different circumstances. Some of the differences are systematic and can be attributed to deeper differences, such as the intention of speaker.

My contribution to this problem is to develop a learning framework that can be used to identify multimodal features (and patterns) that are characteristics of the systematic differences in human conversations and to build automatic detection mechanisms that are robust to spontaneous speech effects. Current projects include automatic topic segmentation and labeling and automatic decision detection. The overarching goal is to provide visual aids at the right level of details for the users to find information from the often-lengthy archives of conversation recordings.

 

Recognition & Awards

 

2017 IBM Manager Choice Award

2016 IBM Innovation Patent Plateau & High-value patent  

2016 IBM Eminence and Excellence Award -- Cognitive Build Finalist Top 4

2016 IBM Technology Leadership Event (TLE), Cognitive Healthcare Challenge lead coach; Quarter-Finalists for the 2017 AI Challeng

2015 Technology Leadership Event: speaker (Disruptive technology) & Manager Choice Award

2014 IBM Manager Choice Award

2013 IBM Invention Achievement Award

2011 IBM Invention Achievement Award

2009 IBM Invention Achievement Award

2007 GOOGLE European Anita Borg Scholar

2005 – 2008 EU FP6 Project: AMIDA (Augmented Multi-party Interaction with Distant

Access)+AMI (Augmented Multi-party Interaction) (FP6-506811)

2004 Winner of Taiwan Merit Scholarship (National Science Council)

2003 Top Scholar Award, University of Washington

 

Community Services

  • 2017    Scientific Program Committee Member, American Medical Informatics Association annual symposium 2017

  • 2017     Program Committee, Workshop for Computational Biology, International Conference on Machine Learning (ICML 2017)

  • 2017    Organizer, IBM Computational Health Summit
  •  2016-17 Secretary, Consumer and Pervasive Health Informatics Work Group (CPHI-WG), American Medical Informatics Association

  • 2016     Advisor, Health Info Lab, Norwegian University of Science and Technology
  •  2016     Chair, AMIA Best Student Paper of Consumer and Pervasive Health award committee

  •  2016     Co-organizer, AMIA Working Group Pre-symposia: Patient-Generated Health Data in Action

  •  2016     Chair, AMIA Annual Symposium Didactic Panel: Transforming Patient-Generated Data for Wellness and Biomedical Research: From Behavioral Sensing to Decision Support

  • Co-chair, IBM Healthcare informatics Professional Interest Group

  • Organizer, IBM Precision Medicine and Wellness Day 2015
  • Managing committee, Emerging Information and Technology Association

  • Board of Director & Treasurer: Chinese Institute of Engineers Greater New York Chapter (CIE-GNYC) 2013-2016
  • 2017     Chair, MEDINFO Panel: Integrating Science of Data with Science of Care for Interpreting Patient Need: Opportunities and Challenges in the New Era of Cognitive Healthcare Solution

  •  2017     Co-Chair, MEDINFO Workshop: From Data Modeling to Knowledge Learning Symbiosis:
The Evolution of Cognitive Data Analysis and Decision Support in Healthcare

  •  2016     Chair, Medical Informatics European / Health – Exploring Complexity (MIE/HEC 2016) Workshop: Interdisciplinary Approaches for Using Visualization for Wellness Decision Support

  •  2016     Chair, MIE/HEC Panel: Putting User-Generated Data in Action: Improving Interpretability for Clinical and Consumer Informatics

  •  2016     Chair, MIE/HEC Workshop: An Socio-Technical Approach to Securing Health Informatics

  •  2015     Chair, MEDINFO Panel: Effective Patient Adherence Management by Engaging Enabling Technologies

    2015     Program Committee, EFMI STC  
  • Invited book chapter: Health Information Management textbook (Springer)

  • Invited speaker: Norweigian University of Science and Technology, US-Taiwan Biotech Business Form, Taipei Medical University-IBM symposium keynote, TsingHua University X-health Lab, Columbia U Roadmap, Columbia U Big Data Forum, International Chinese Statistical Association Annual Meeting (ICSA) 2017 

  • Invited speaker: IBM Technology Leadership Event (TLE) -- disruptive technologies

  • Invited Session Chairs: Applied Human Factors and Ergonomics Conference (AHFE), Industrial and Systems Engineering Research Conference (ISERC), IEEE International Conference of Service Operations, Logistics and Informatics (IEEE SOLI), IEEE CollaborateCom Healthcare, CIE-GNYC conference (Healthcare session)

  • Workshop Organization: MEDINFO 2017; AMIA 2016; MEDINFO  2015, MIE  2015: Effective adherence management with exogenous data analytics

  • Panel Organization MEDINFO 2013: Personalized healthcare and management: potentials and challenges

  • Workshop Organization MIE 2014: gaps analysis of patient-controlled devices

  • Organizing Committee for the Standardization Work Group on Data for Science and Technology: Chronic Disease Management and Independent Living for the Aged (2011- present)
  • IBM Health Care and Life Science (HCLS) Webinar Series Organizer
  • IBM Academy of Technology Conference Committee
  • IBM HCI PIC Research Coordinator (2010-present)
  • IBM Westchester Toastmaster Club, President(2011-present), Secretary (2010), Sergeant-at-arm (2009)
  • UC Berkeley Alumni Society
  • PC: Annual Conference of Human Language Technology (HLT), European Association of Computational Linguistics (EACL), North American Association of Computational Linguistics (NAACL)
  • Journal review: IEEE Intelligent Systems Transactions on Knowledge and Data Engineering Statistical Analysis and Data Mining IEEE Journal of Selected Topics in Signal Processing Journal of Natural Language Engineering
  • Conference paper review: NAACL, EACL, HLT, ICML, ICIS, ACL, CHI
  • Statistical Natural Language Processing Reading Group
  • Women in Machine Learning Workshop (WiML)
  • EUROMASTERS Summer School in Speech Technology
  • IGK Summer School in Computational Linguistics and Psycholinguistics, Univ. of Saarland
  • European Summer School in Logic, Language, and Information (ESSLII)