Health Informatics       

links

Health Informatics - Talks


Watson Healthcare Informatics PIC Seminar Series

Watson HI PIC Chair: 
Pei-Yun (Sabrina) Hsueh (phsueh@us.ibm.com)

2018 TALK SCHEDULE

 

DATE: DECEMBER 7, 2018, 1:00PM - 2:00PM

LOCATION: YKT 20-043

SPEAKER: Dr. Geoffrey Siwo, Assistant Professor, Department of Biological Sciences, University of Notre Dame (https://biology.nd.edu/people/geoffrey-siwo/)

TITLE: Accelerating knowledge discovery and equity in genome editing

SUMMARY: Genome editing technologies such as CRISPR/Cas9 are transforming fundamental biomedical research, could enable new treatments for animal and plant diseases, and open new ways for engineering biological organisms for food and industrial applications. Therefore, a global understanding of the field is needed to identify challenges, risks, opportunities and biases that could shape the impact of the technology. In addition, to keep pace with the rapid advancements in genome editing technologies, automated approaches for knowledge discovery are needed. In this talk, I will present our work using natural language processing and machine learning methods to track the global state of genome editing technologies based on scientific literature and the global supply of DNA. Our work identifies emerging biases, potential security risks, funding patterns and the potential for automated scientific discovery from literature in informing genome editing research.


DATE: NOVEMBER 19, 2018, 11:00am - noon EST

LOCATION: Cambridge: 2T08

SPEAKER: Kenney Ng, IBM Research

TITLE: IBM Health Corps and CARE team to bring data-driven decisions for health management to Bihar India

SUMMARY: CARE is an international non-governmental organization (NGO) that supports 94 countries on poverty-fighting development and humanitarian aid projects. IBM and CARE collaborated on an IBM Health Corps initiative to help health services delivery in Bihar, India where 55% of the population lives below the poverty line, and high infant and maternal mortality rates persist. The team designed and built an interactive dashboard integrating multiple healthcare systems to help public health managers make more informed and effective decisions for health care delivery in the region. Seven IBMers from Research (WRC, CRL), Watson Health, Watson Analytics, and GBS (India) conducted a 3 week project that focused on creating an extensible data integration and analytics platform and using it to demonstrate value on several drug supply chain use cases. The platform and proof of concept ( https://youtu.be/19vP-9legCQ ) was delivered 09/2018 to CARE India, which is working with the Bihar state government to enhance and expand it. This talk will describe the project and experiences of the IBM Health Corp team ( https://ibmhealthcorpsblog.wordpress.com ).

BIO: Kenney Ng is currently a research staff member in the Center for Computational Health and manager of the Health Analytics Research Group at IBM Research Cambridge. His current research focus is on the development and application of data mining, machine learning, and AI techniques to analyze, model and derive actionable insights from real world health data. His prior research areas include information retrieval, speech recognition, probabilistic modeling, topic modeling, and statistical language modeling. Before IBM Research, Kenney was a senior software engineer and architect in IBM Software Group. Prior to IBM, Kenney was a principal software engineer at iPhrase Technologies and held research positions at the MIT Laboratory for Computer Science, BBN Technologies, and MIT Lincoln Laboratory. Kenney holds Bachelors, Masters, and Ph.D. degrees in Electrical Engineering and Computer Science from the Massachusetts Institute of Technology. He is a member of IEEE and AMIA.
Kenney Ng (kenney.ng@us.ibm.com)

LINKS: Calendar enrollment: https://learn.atlanta.ibm.com/ec/session.html?id=A4724DD9A909B9EF852583360074A4E2&action=enroll
Note that this talk will be in Cambridge and broadcast via webex: Webex: https://ibm.webex.com/meet/kenney.ng


DATE: NOVEMBER 12, 2018, 1:00-2:00pm

LOCATION: YKT 20-001

SPEAKER: Dr. Shabbir Syed Abdul (Associate Professor, Graduate Institute of Biomedical Informatics, Taipei Medical University)

TITLE: Transforming healthcare from Evidence-based to Intelligence-based Medicine

SUMMARY: Artificial intelligence (AI) is a discipline of computer science where software is created that emulates human intelligence. Artificial Intelligence (AI) is increasingly being used in health care. AI is being used in diagnosis, therapy, automatic classification, and rehabilitation. Machine learning (ML) is one application of AI which is being implemented in order to find hidden information from available data. There are many ML algorithms that have been developed classified as supervised learning, unsupervised learning, ensemble learning and reinforcement learning. This presentation focus on four major topics such as participatory health, prevention care, Medical errors and personalized prevention.

BIO: Dr.Shabbir Syed-Abdul is an associate professor at Institute of Biomedical Informatics, a leading researcher and a principal investigator at the International Center for Health Information Technology, Taipei, Taiwan. His major research interests are Long-term care with Wearable technologies, mHealth, Big data analysis and visualisation, Artificial Intelligence, Personal Health Records, Social Network in healthcare and Hospital Information System. He wants to empower care providers and improve patient engagement. He feels one of the ways to achieve it is to focus on the management and flow of the health/medical information among health care providers and seekers.

LINKS: Event Enrollment: https://learn.atlanta.ibm.com/ec/session.html?id=0DC4218354D84B97852583360039C3DD&action=enroll

 


DATE: OCTOBER 22, 2018, 11am - noon

LOCATION: YKT 20-001

SPEAKER: Marissa Burgermaster, PhD, Asistant Professor at University of Texas at Austin

TITLE: Precision Behavioral Nutrition: Improving Nutrition Interventions with Innovative Technologies

SUMMARY: When faced with chronic disease, patients are often inspired or encouraged to change their diet. However, adopting and maintaining a new diet is challenging for many and behavioral nutrition interventions typically have small effect sizes. This is at least partly because individuals respond differently to interventions. In this talk, Dr. Burgermaster will present how she has applied informatics methods, including knowledge engineering, data science, and human computer interaction, in three projects that address different aspects of precision behavioral nutrition.

BIO: Dr. Marissa Burgermaster is an assistant professor at University of Texas at Austin. She is joint appointed to the faculties of Nutritional Sciences in the College of Natural Science and Population Health at Dell Medical School. Her research applies data science and human-computer interaction methods to improve nutrition and community health.

Prior to joining UT, Dr. Burgermaster completed a postdoctoral fellowship in biomedical informatics at Columbia University Irving Medical Center where she conducted research on technologies for diabetes management among underserved New Yorkers. She was granted an early career award from the Sackler Institute for Nutrition Science at the New York Academy of Sciences to support her work developing methods for psychosocial phenotyping. She was also the behavioral nutrition lead for the CUIMC-based team that developed “Taming Type 2 Diabetes Together (T2D2),” a voice application for personalized nutrition and diabetes self-management that was a finalist in the 2017 Alexa Diabetes Challenge and the 2017 World Cup of Voice Tech in Diabetes. Dr. Burgermaster holds a PhD in Behavioral Nutrition from Columbia University, where her research at Teachers College’s Tisch Center for Food, Education, & Policy focused on improving the evaluation of school-based childhood obesity prevention interventions. Dr. Burgermaster holds an MS in Nutrition and Food Science from Montclair State University and an MAEd in Curriculum and Instruction from William & Mary. She previously had a career as a teacher and school administrator.

LINKS: Video streaming: https://watmedia.res.ibm.com/Mediasite/Play/1a3e42129a5f40dc9a6f73ea1a6f79ad1d

Calendar Invite: https://learn.atlanta.ibm.com/ec/session.html?id=14B86C43ED4185B88525831E0036AD09

Event Enrollment: https://ec.w3bmix.ibm.com/event.html?id=F47EBBEFD1F011D58525831E0036A9B7


DATE: OCTOBER 19, 2018, 11:00am - noon

LOCATION: YKT 20-043

SPEAKER: Lena Mamykina, PhD; Columbia University

TITLE: From personal informatics to personal analytics: personalized decision-support in health

SUMMARY: The increasing abundance of personal data related to health and wellness presents new opportunities for discovery and insight and can help individuals learn from their own experiences, as well as from experiences of others. These trends inspired active research in machine learning and data mining; they also present new opportunities for research in interactive systems. There remain many open questions as to how to design interactive solutions that leverage new streams of personal and social data and new data science capabilities to promote self-management of chronic diseases. In my research, I investigate these questions in the context of self-management of type 2 diabetes, and, specifically, nutrition management. In this talk I will discuss several ongoing research initiatives that strive to help individuals make informed choices by promoting reflection, providing decision-support, and generating personalized recommendations.

BIO: Dr. Lena Mamykina is a Florence Irving Assistant Professor of Biomedical Informatics at the Department of Biomedical Informatics at Columbia University. Dr. Mamykina’s research resides in the areas of Biomedical Informatics, Human-Computer Interaction, Ubiquitous and Pervasive Computing, and Computer-Supported Collaborative Work. Her broad research interests include individual and collective cognition, sensemaking, and problem-solving in the context of health and wellness. She is specifically interested in novel interactive solutions that take advantage of new streams of personal and social data and novel data science capabilities. Dr. Mamykina received her B.S. in Computer Science from the Ukrainian State University of Maritime Technology, M.S. in Human Computer Interaction from the Georgia Institute of Technology, Ph.D. in Human-Centered Computing from the Georgia Institute of Technology, and M.A. in Biomedical Informatics from Columbia University. Her dissertation work at Georgia Tech focused on facilitating reflection and learning in context of diabetes management with mobile and ubiquitous computing. Prior to joining DBMI as a faculty member, she completed a National Library of Medicine Post-Doctoral Fellowship at the department. 

LINKS: Video streaming: https://watmedia.res.ibm.com/Mediasite/Play/d0765a6282124c99bca26788842cfe421d

Calendar Invite: https://w3-connections.ibm.com/communities/service/html/communityview?communityUuid=03b327b5-4902-4bbf-9daa-e57c0e16e29b#fullpageWidgetId=W9708462ff71f_4dbb_b8d5_2afa4db602e6&eventInstUuid=bf4f9f8a-979e-4d79-bfe2-016b752645bc

Event Enrollment: https://ec.w3bmix.ibm.com/event.html?id=7590FC853D7994FB8525831E003AE6D3


DATE: OCTOBER 4, 2018

LOCATION: YKT 20-001

SPEAKER: Richard C. Liang, Research Assistant Professor University of Michigan Medical School, Department of Neurology

TITLE: Define the phenotypic modifiers of DYT1 dystonia

SUMMARY: Dystonia is a neurological disorder characterized by debilitating prolonged involuntary movements. The movement deficits can manifest as an isolated symptom or result from pathological insults (e.g., head trauma and antipsychotic medication) to the central nervous system. DYT1 dystonia, the most common form of the primary dystonia affecting approximately 30,000 persons in the United States, is an inherited disease caused by a dominant deletion mutation in the TOR1A gene. ~30% of the mutation carriers develop motor deficits, ranging from mild features to profound dystonia. The highly variable phenotypic expression of this mutation, even between siblings, suggests additional genetic background influencing the disease penetrance. At present, there is no effective intervention for DYT1 dystonia, which results in increased medical costs and frustration among patients and providers. Strategies focusing on the diagnosis to predict the risk probability of dystonia development and novel therapeutics for the disease are urgently needed for preventive interventions, slowing the progression, and precision treatments. Little is known about what molecular mechanisms involved in the DYT1 pathogenesis. Recent in-vivo and in-vitro murine studies show that DYT1 mutation leads to ubiquitin aggregates in neurons and neurodegeneration. Deletion of Tor1b, a paralog of Tor1a, worsens the neuropathology on the DYT1 mutation background but generates zero impact on the overall health of the animals when both Tor1a alleles are normal. In this case, the offspring manifesting-carrier receives genes - the DYT1 mutant and a Tor1b knockout allele -from the non-manifesting carrier and the non-carrier parents, respectively. Therefore, Tor1b is identified as a phenotypic modifier because its deficiency only increases the cumulative risk of developing DYT1 dystonia. Additionally, the 1000 Genomes Project discovers that human TOR1B has two stop-gained variants (rs536099033 and rs377407558) and three frameshift variants, suggesting the threat of TOR1B deficiency in manifesting DYT1 dystonia. Given that a typical DYT1 patient's family including manifesting carrier(s), non-manifesting carrier(s), and non-carrier(s), it is likely to highlight the phenotypic modifiers shared exclusively between the manifesting carriers and their non-carrier parents through the genomic comparisons. Our strategy will compare the DNA data from the newly built DYT1-family Trio genomic database to define the genetic components that contribute to the symptom-onset of DYT1 dystonia. Each potential modifier candidates will be validated by its influential power on the DYT1 mutation expression using our pre-clinical models. The outcome may develop the mathematical formula that accurately predicts the penetrance probability and the age-onset for a given DYT1 mutation carrier, which can assist the primary care doctors in providing the early surveillance and the suitable preventive interventions to the patients.

BIO: Richard Liang’s career focuses on human disease modeling to decipher the molecular mechanisms of pathogenesis. At present, his research direction is to develop the interventions of childhood-onset dystonia. He is collaborating with the Dystonia Medical Research Foundation and the bioinformaticians to identify the biomarkers and the therapeutic targets for providing the early surveillance and the suitable preventive interventions to the patients. His early scientific achievements include the establishment of the in vitro scratch assay for cancer research, the generation of the first dystonia animal model, and the discovery of a novel mechanism driving developmental neurodegeneration. He is a current editorial board member of Scientific Reports.

LINKS: Calendar invite: https://ec.w3bmix.ibm.com/session.html?id=DD2927DD75EC343F852582F6004F9874&action=enroll

Video streaming: https://watkvideo01.watson.ibm.com/Mediasite/Play/2b55b36f12d54192933d15fd0fc08e741d


DATE: SEPTEMBER 12, 2018, 3:00-4:00pm

LOCATION: YKT 20-043

SPEAKER: Dr. Timothy Buchman, Director, Emory Critical Care Center, Professor of Surgery, Anesthesiology, and Biomedical Informatics, Emory University School of Medicine; Editor-in-Chief, Critical Care Medicine

TITLE: The Future of Machine Learning and AI in Clinical Critical Care

SUMMARY: Risk-stratification and prediction in critical care historically relied on one-time assessments of initial conditions. With the appearance of electronic medical records and improved availability of near-continuous, high resolution data, attention has shifted to repeated analysis of physiologic trajectory, including the interventions aimed at altering those trajectories. In this presentation, Dr. Buchman will discuss the evolution of predictive tools in the context of sepsis, with specific attention to artificial intelligence grounded in machine learning and deep learning.

BIO: Dr. Timothy G. Buchman, Ph.D., M.D., FACS, FCCM serves as Director of Emory Critical Care Center at Emory University Hospital Midtown. Dr. Buchman serves as a Director of Emory University Hospital Midtown. Dr. Buchman serves as Member of Scientific Advisory Board at Therapeutic Monitoring Systems Inc. He served as Director of Emory Critical Care Center and Director at Emory Healthcare, Inc. Before joining Emory, he served as professor of surgery and director of Acute and Critical Care Surgery at Washington University School of Medicine in St. Louis. Prior to his 15 years on the faculty at Washington University, Dr. Buchman directed the surgical intensive care unit and the trauma center at Johns Hopkins Hospital in Baltimore, where he completed his surgical training.

LINKS: Calendar Invite: https://ec.w3bmix.ibm.com/session.html?id=E5D14FBBF77AA59C8525830100613C97


 

DATE: JULY 19, 2018

LOCATION: YKT 20-043

SPEAKER: Charles Senteio, Assistant Professor, Department of Library and Information Science, Rutgers School of Communication and Information

TITLE: Participatory Design to help address barriers to technology for chronic disease self-management

SUMMARY: In this talk Prof. Senteio will discuss barriers, and potential facilitators, to health information exchange which can inform clinical decisions to help address persistent chronic disease disparities. Among known drivers of chronic disease disparities are barriers to information exchange between chronic disease patients and clinicians. Patient-specific health information, specifically psychosocial information, informs clinical decisions. Generally, primary care information flows and EHRs do not enable clinicians to consistently and methodically collect and use psychosocial information, “there’s no tab for poverty”. For example, EHR capabilities can inform clinicians that a medication has not been filled, but they do not necessarily tell them why. Awareness of psychosocial information can provide insight for barriers to recommended chronic disease self-management (i.e., medication behavior, diet, physical activity, attendance at follow-up appointments), and subsequently can inform clinical decisions to address them. Consumer-oriented technologies (i.e., patient portals, smartphone apps, smartwatches, etc.) can enable the capture and use of health information to inform facilitators and barriers, however; specific patient groups experience barriers to these technologies (i.e., elders, low SES/Heath Literacy, minorities). Barriers to technology use result in intervention-generated inequalities (IGIs) – developers may be exacerbating disparities in access and use – which may contribute to chronic disease disparities. Health informatics developers and researchers should consider IGIs, particularly given that patient groups that experience barriers to technology also experience chronic disease disparities for incidence, complications, co-morbidity, and mortality. As a community health informatics researcher focused on how technology can help facilitate information exchange, I will discuss the potential in the development of technology-enabled capabilities using participatory design (PD). Specifically, I am interested in using PD with emerging technology (i.e., case based reasoning, virtual reality) to support health information exchange between providers and patients.

BIO: Charles Senteio is a health informatics researcher focused on improving chronic care outcomes. His is an Assistant Professor at the Rutgers School of Communication and Information in the Department of Library and Information Science. He uses mixed methods to investigate how healthcare practitioners and patients can better use information to improve chronic disease outcomes for at-risk patients – while reducing cost of care – through financially sustainable care delivery models. He develops and enhances innovative, scalable approaches to care delivery, with a particular emphasis on community-based participatory (CBPR) research strategies. In 2015 he received a PhD in health informatics from the School of Information at the University of Michigan, and also completed a Masters in Social Work during the doctoral program. He began his IT strategy consulting career in the high-tech industry after receiving an MBA from the Ross School of Business at the University of Michigan, and started a healthcare consulting practice in 2002 to concentrate exclusively on healthcare. He is a licensed master social worker (LMSW), certified health education specialist (CHES) and certified community health worker instructor (CHW-I

LINKS: Video streaming: https://watkvideo01.watson.ibm.com/Mediasite/Play/55c6d0f623ea42c9aa5599a6c0024fef1d


 

DATE: FEBUARY 27, 2018

LOCATION: YKT 20-043

SPEAKER: Nicholas J Belkin, Distinguished Professor of Information Science, Department of Library & Information Science, Rutgers University

TITLE: Exploring Information Seeking and Searching Intentions: An Overview of Recent Research in Personalized Search at Rutgers University

SUMMARY: In its broadest sense, information retrieval (IR) is concerned with helping people to achieve goals, or accomplish tasks, by supporting useful interactions with information systems and information objects. The kinds of interactions that IR supports are those associated with information seeking and information searching, with the eventual goal being that the person who initiates an information seeking/search session find information that is useful for achieving the goal, or accomplishing the task, that motivated the search session. It has been demonstrated that different types of goals or tasks lead to quite different information seeking and information searching behaviors. Additionally, individual characteristics of the person, and aspects of the person’s context and situation other than the motivating task, affect their behaviors during information seeking and searching. These factors motivate the desire to personalize the support offered by information retrieval systems; the most common approach to personalization has been to identify information objects likely to be of use to the specific person. However, it is the case that persons often engage in extended information searching sessions, during which they engage in a variety of different behaviors, motivated by different intentions, such as learning about a domain, deciding on good search terms, evaluating information objects, comparing information objects, finding a specific information object, and so on. Each such intention leads to different kinds of interactions (or at least, the desire on the part of the searcher for a different kind of interaction) that would be most appropriate for accomplishing the intention. Understanding information seeking and searching in this way suggests that it is important for IR systems to be able to identify the persons' intentions during the course of an information searching session, in order to personalize the forms and means of interaction with both the IR system and the information objects within that system, to those specific search intentions. At Rutgers University, we have been engaged in a long-term research program in identifying, during the course of an information search session, people’s motivating goals, their individual characteristics, and their information search intentions, solely through the person's behaviors during that session. In this talk, I expand upon the motivation for this approach to the design of IR systems, and survey some of our major results in identifying and understanding information seeking and searching intentions, with specific attention to personalization at the within search session level.

BIO: Nicholas Belkin is Distinguished Professor of Information Science in the Department of Library & Information Science, Rutgers University. Previous to that appointment, he was at the Department of Information Science, The City University, London. He has held visiting positions at the University of Western Ontario, the Free University, Berlin, GMD-IPSI, Darmstadt, Germany, and the Institute for Systems Science, National University of Singapore, as well as Fulbright Fellowships in Finland and Croatia.

Professor Belkin has served as the Chair of the ACM SIGIR, and President of the Association for Information Science and Technology (ASIST). He is the recipient of the ASIST’s Outstanding Teacher award, its Research Award, and its Award of Merit. In 2015, he received the ACM SIGIR Gerard Salton Award, for significant, sustained and continuing contributions to research in information retrieval.

Professor Belkin is known as one of the founders of the cognitive viewpoint in information science, and as a leader in integrating information behavior research with information retrieval research. His most recent research has focused on personalization of interaction with information, especially with respect to the nature of the task which leads people to engage in information seeking, and on methods for evaluation of whole-session search. His current research project, Characterizing and Evaluating Whole Session Interactive Information Retrieval, is supported by the US National Science Foundation.

LINKS: Video Steaming: https://watkvideo01.watson.ibm.com/Mediasite/Play/a186849b911c43368a827ccbfbb9c3d31d


 

DATE: FEBUARY 5, 2018; Talk: 11-noon; Method forum: 1-2pm

SPEAKER: Professor Rupa Valdez, University of Virginia

TITLE: Patient Work As Design Foundation: The Case of Health Information Communication

SUMMARY: The home and community are increasingly becoming prominent spaces of health management as the locus of care shifts away from institutional settings. Consumer health IT intended to support patients with health management in these spaces is rapidly proliferating and includes innovations such as personal health records, decision support systems, and online health communities. Despite the promise of these solutions, patients’ adoption of consumer health IT has remained low. Evidence suggests that lower adoption rates stem in part from a failure to design consumer health IT in a way that is aligned with the realities of patients’ everyday lives. Designing technology that meets patients’ needs requires studying their health-related activities (i.e. patient work) and the context in which these activities occur (i.e. patient work system). In this talk, Dr. Valdez will discuss three studies focused on understanding patients' work of communicating health information to their social network members and the resulting implications for consumer health IT design. The first study focuses on individuals seeking care at federally qualified health centers, the second on type 2 diabetes patients who use Facebook, and the third on individuals living with disabilities.

BIO: Dr. Valdez received her PhD at the University of Wisconsin in 2012. She is a human factors engineer and informaticist who has conducted research and lectured on topics at the intersection of consumer health IT, human factors engineering, public health, and cultural anthropology. Her work has a strong focus on underserved populations, including racial/ethnic minorities and individuals living with disabilities. Dr. Valdez has received research support from multiple federal organizations, including the National Institutes of Health, the Agency for Healthcare Research and Quality (AHRQ), and the National Science Foundation. She has also provided consulting services to the National Academies of Science, the Veterans Health Administration, and AHRQ. She served on the steering committee for AMIA's Health Policy Meeting on Advancing Patient Centered Care, Collaboration, Communication, and Coordination and as co-chair for the Health IT Track of the 2014 International Symposium on Human Factors and Ergonomics in Health Care. Dr. Valdez currently serves as an Associate Editor of JAMIA Open. She is an assistant professor of Biomedical Informatics at the University of Virginia.

LINKS: Video streaming: https://watkvideo01.watson.ibm.com/Mediasite/Play/251f440338bf4f759a8c41b648c569321d
Any questions you would like to ask the speaker, please email/ping Sabrina
(There will be no video streaming for the methodology forum)Sign up for 1:1 meeting on wiki: https://goo.gl/P9fb9z
or contact host: Pei-Yun Sabrina Hsueh (phsueh@us.ibm.com)


 

DATE: JANUARY 18, 2018

LOCATION: YKT 20-043

SPEAKER: Professor Jimeng Sun, Georgia Tech

TITLE: Doctor AI - Interpretable Deep Learning Methods for modeling Electronic Health Records

SUMMARY: Deep neural networks provide great potential to create better models for longitudinal electronic health records (EHRs). In this talk, we will present a series of case studies of deep learning examples for modeling EHR.
1) We illustrate how recurrent neural networks (RNN) can be used to model temporal relations among events in electronic health records (EHRs) to predict heart failures.
2) We present Med2Vec, which not only learns the representations for both medical codes and visits from large EHR datasets with over million visits, but also allows us to interpret the learned representations confirmed positively by clinical experts.
3) We introduce an interpretable predictive model RETAIN which achieves high accuracy while remaining clinically interpretable and is based on a two-level neural attention model that detects influential past visits and significant clinical variables within those visits (e.g. key diagnoses).
4) We propose GRaph-based Attention Model (GRAM) that learn the clinically interpretable representation from both EHRs and medical ontologies.
5) Finally, we present a new approach, medical Generative Adversarial Network (medGAN), to generate realistic synthetic patient records.

BIO: Jimeng Sun is an Associate Professor of College of Computing at Georgia Tech. Prior to Georgia Tech, he was a researcher at IBM TJ Watson Research Center. His research focuses on health analytics and machine learning, especially in designing tensor factorizations, deep learning methods, and large-scale predictive modeling systems. He published over 120 papers and filed over 20 patents (5 granted). He has received SDM/IBM early career research award 2017, ICDM best research paper award in 2008, SDM best research paper award in 2007, and KDD Dissertation runner-up award in 2008. Dr. Sun received B.S. and M.Phil. in Computer Science from Hong Kong University of Science and Technology in 2002 and 2003, M.Sc and PhD in Computer Science from Carnegie Mellon University in 2006 and 2007.

 

 

2017 TALK SCHEDULE
Date Speaker Time & Location Title and additional info Link to Video Streaming, Slides and Replays
Jan 9
Sudy Majd, Columbia University
11am (20-001)

Talk Title: Tools to Assist Restrained Eaters: A Query Theory Approach

Abstract: At any given moment, approximately 45 million Americans are continuously practicing some form of dietary restraint in order to control or reduce body weight. Despite their best efforts, these individuals, known as Restrained Eaters, rarely reach their goals. In the last two decades, there has been a surge of research aimed at understanding why Restrained Eaters fail to reach their goals but with less success at understanding how Restrained Eaters can reach those goals. Behavioral interventions designed to influence behaviors at the very moment of decision and temptation have shown to be the most effective at getting people to choose healthier options. In this talk, I present research findings that extend recent thinking in the field of decision-making and behavioral economics to Restrained Eaters. Specifically, I test the usefulness of Query Theory as a behavioral intervention tool to change the decisions of Restrained Eaters through subtle manipulations to the decision process. Query Theory has been widely used to change the decision of individuals in a variety of choice scenarios, but this research is the first to use it to change food decisions.

Bio: Sudy Majd is a doctoral candidate in the Psychology Department at Columbia University. Her research applies behavioral models of decision making and motivation science to food choices and health behaviors. She believes behavioral interventions paired with psychological theories are an underutilized approach to improving health behaviors. Prior to her doctoral career, Majd spent several years working for a retail consulting firm researching in-store and online consumer behavior around the world. She also holds a MS in Applied Research from the department of Design and Environmental Analysis at Cornell University.

 

 

 

Feb. 1
Patricia F. Brennan, RN, PhD, MS



Director, US National Library of Medicine
11 am (Auditorium)  

Talk Title: Anticipating the National Library of Medicine's 3rd Century'

Abstract: NLM is poised to become the epicenter of data science for NIH.  Dr. Brennan will talk about how she sees the NLM supporting the transformation of data to information and then to knowledge to support health care information needs of the present and of the future.

Brief Bio: Dr. Brennan came to NLM in August 2016 from the University of Wisconsin-Madison, where she was a professor at the School of Nursing and College of Engineering.  She is a pioneer in the development of innovative information systems and services, such as ComputerLink, an electronic network designed to improve the lives of home care patients and increase their independence.  She directed HeartCare, a web-based information service that helps home-dwelling cardiac patients recover faster, with fewer symptoms, and Project Health Design, an initiative designed to stimulate the next generation of personal health records.

A past president of the American Medical Informatics Association, Dr. Brennan was elected to the Institute of Medicine in 2001.  She is a fellow of the American Academy of Nursing, the American College of Medical Informatics, and the New York Academy of Medicine.

 

 

 

 

Feb 10
Wendy Chapman ( chair of the University of Utah, Department of Biomedical Informatics ) 11 am (20-001)

Talk Title: What do you mean when you say you want to find patients with a cough? Knowledge representation to support phenotyping from text

Abstract: Leveraging clinical narratives to classify patients based on phenotype requires layers of annotations. Representation of the knowledge described in the reports is critical to accurate extraction of that information. In this talk, Dr. Chapman will describe application ontologies her lab has developed for modeling annotations of information described in clinical reports. She will illustrate the usefulness of the formalism with several use cases and describe a vision of how the ontologies can potentially support collaborative knowledge authoring and NLP customization.

Bio: Dr. Chapman earned her Bachelor's degree in Linguistics and her PhD in Medical Informatics from the University of Utah in 2000. From 2000-2010 she was a National Library of Medicine postdoctoral fellow and then a faculty member at the University of Pittsburgh. She joined the Division of Biomedical Informatics at the University of California, San Diego in 2010. In 2013, Dr. Chapman became the chair of the University of Utah, Department of Biomedical Informatics.

Dr. Chapman’s research focuses on developing and disseminating resources for modeling and understanding information described in narrative clinical reports. She is interested not only in better algorithms for extracting information out of clinical text through natural language processing (NLP) but also in generating resources for improving the NLP development process (such as shareable annotations and open source toolkits) and in developing user applications to help non-NLP experts apply NLP in informatics-based tasks like clinical research and decision support.

Prof. Chapman's presentation deck on Box: https://ibm.box.com/s/6nsp35mukeg4afn86gg36ke55dgtmlpg

Smart meeting recording on Box:

https://ibm.box.com/s/tr3n60ulro9pjaovgznhppkk2el2w45v

Access to audio recording:

Recording Conference URL:        https://www.teleconference.att.com/

conferencerecord

Recording ID:        1486742408

Participant code: 3169-7157

Feb 24 Avi Yaeli
 
1pm (20-001)

Talk Title: Visual Analytics for Data Science in Healthcare 

Abstract: The growing availability of large healthcare observational data, instrumented device data and other information sources coupled with the power of advanced analytics and cloud is expected to transform the healthcare industry, leading to better health and care, and at lower costs.   Visual Analytics can be described as the use of interactive visual interfaces to drive knowledge discovery and support decision making tasks performed by humans.  In this talk I will present an overview of visual analytics, its role and opportunity in supporting cognitive computing in healthcare.  I will present and demo several technologies we developed for interactive visual data exploration and understanding of healthcare data and the way we integrate with Jupyter Notebooks in order to support data scientists and clinical researchers working in this environment.

Bio: Avi Yaeli joined IBM in Research in 1998 as a Research Staff Member.   In 2016 he joined Watson Health Innovation to develop and apply the practices of visual analytics in Healthcare and Life Sciences applications.  Previously he was involved in research and development activities in the areas of location-aware systems, spatiotemporal data mining, anomaly detection, information visualization, open linked data, and smarter cities solution. Mr. Yaeli has published more than 30 papers and patents and serves as IBM Master Inventor.

 

 

March 1 Yize Zhao, Department of Healthcare Policy and Research, Weill Cornell Medicine 11am-noon (20-001)

Talk Title: Bayesian Feature Selection for Ultra-high Dimensional Imaging Genetics Data

Abstract: This work is motivated by the joint analysis of multivariate phenotypes and ultra-high dimensional genotypes obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI). Currently, such joint analysis presents major computational and theoretical challenges for existing statistical methods. The aim of this work is to propose a novel multilevel sequential selection procedure under a Bayesian multivariate response regression model (MRRM) to select informative features among multivariate responses and ultra high-dimensional predictors. Specifically, we treat the identification of nonzero elements in the sparse coefficient matrix as a hierarchical feature selection problem by first selecting potential nonzero rows among the matrix (genotype selection) and then localizing the nonzero elements within the marked rows (phenotype selection).  The genotype-wise selection is accomplished by constructing multilevel auxiliary selection models under different scales with the actual scale auxiliary model treated as another level for the ultimate phenotype-wise selection. This procedure allows the posterior inference be "reweighted'" to concentrate more efficiently on the potential signals in a sequential fashion, which dramatically reduces the computational cost and improves the mixing of Markov chain. Extensive simulations are provided to show the superiority of our method compared with several completing approaches. We also apply the method to the ADNI data with biologically meaningful results obtained. 

Short bio: Yize Zhao is an Assistant Professor in Biostatistics from Department of Healthcare Policy and Research, Weill Cornell Medicine, Cornell University. Her methodology research focuses on the development of statistical methods, in particularly scalable Bayesian approaches to analyze large-scale complex data to achieve feature selection, prediction, data integration and network analysis. She has strong interests in applications on imaging (fMRI, PET and DTI), imaging genetics and omics data, and she recently also starts to work on statistical method development and data analyses for electronic health record data and lifetime science data. 

Video recording:

http://watkvideo01.watson.ibm.com/Mediasite/Play/20991fb958b944ec92f98e474726116b1d

March 7 David C. Mohr, Ph.D,  professor in the Northwestern University Feinberg School of Medicine’s Departments of Preventive Medicine, Psychiatry, and Medical Social Sciences, and the Director of Northwestern University’s Center for Behavioral Intervention Technologies 2-3pm (05-235)

Talk Title: Digital Mental Health for Depression and Anxiety

Abstract: This talk will provide a brief overview of digital mental health, which has used common technologies such as web and mobile apps to support behaviors aimed at reducing mental health symptoms. While the potential effectiveness of these approaches has been demonstrated in more than 50 randomized controlled trials, attempts to implement these tools in healthcare settings have failed. Patients do not use the tools, care managers do not know how to support the technologies, and there is no clear implementation strategy.  This talk will present the IntelliCare platform, which aims to provide a more flexible design to meet patients’ needs.  Future directions to decrease patient burden and increase scalability will be presented for discussion, including developing a recommender system for mental health apps, personal sensing using phone sensors, and provision of support for care managers to improve processes of care. 

Link to slides: https://ibm.box.com/s/qjglxpdyvh9yy3gx8gvzkn37r80sgwhi

 

March 13
David Madigan
Columbia University
(joint work with Martijn J. Schuemie, Patrick B. Ryan, George Hripcsak, and Marc A. Suchard)
1pm (20-043)

Talk Title: Honest learning for the healthcare system: large-scale evidence from real-world data

Abstract: In practice,  our learning healthcare system relies primarily on observational studies generating one effect estimate at a time using customized study designs with unknown operating characteristics and publishing -- or not  -- one estimate at a time. When we investigate the distribution of estimates that this process has produced, we see clear evidence of its shortcomings, including an over-abundance of estimates where the confidence interval does not include one (i.e. statistically significant effects) and strong indicators of publication bias. In essence, published observational research represents unabashed data fishing. We propose a standardized process for performing observational research that can be evaluated, calibrated and applied at scale to generate a more reliable and complete evidence base than previously possible, fostering a truly learning healthcare system. We demonstrate this new paradigm by generating evidence about all pairwise comparisons of treatments for depression for a relevant set of health outcomes using four large US insurance claims databases. In total, we estimate 17,718 hazard ratios, each using a comparative effectiveness study design and propensity score stratification on par with current state-of-the-art, albeit one-off, observational studies. Moreover, the process enables us to employ negative and positive controls to evaluate and calibrate estimates ensuring, for example, that the 95% confidence interval includes the true effect size approximately 95% of time. The result set consistently reflects current established knowledge where known, and its distribution shows no evidence of the faults of the current process. Doctors, regulators, and other medical decision makers can potentially improve patient-care by making well-informed decisions based on this evidence, and every treatment a patient receives becomes the basis for further evidence.

Slides available:

Madigan OHDSI IBM March13 2017.pptx

April 3

Sharon Hensley Alford,

Watson Health

11am-noon

(20-001)

Talk Title: Novel Approach to Cancer Biomarkers

Abstract:​ A new approach to cancer screening is needed. Available methods have been called into question regarding efficacy, cost, and patient acceptability. As data on the individual increases, new approaches could offer benefits for each of these difficulties. Dr. Alford will discuss how a new direction in cancer screening and early detection may be in reach.

Bio: Sharon Hensley Alford is a cancer epidemiologist who joined Watson Health in 2015. Dr. Alford has 15 years of experience working with clinical, claims, molecular, and patient reported data for academic and industry motivated research. Dr. Alford is interested in cancer biomarkers for the screening and early detection of cancer.

Slides available:

Novel Approach to Cancer Biomarkers.pptx

May 16

Prof. Veljko Milutinovic (School of Electrical Engineering, University of Belgrade)

 

Life Member of the ACM, Fellow Member of the IEEE, Member of Academia Europaea, Member of the Serbian Academy of Engineering

Member of the Advisory Board of the Vienna Congress COMSULT

Member of the Scientific Advisory Board of Maxeler Technologies

2-3pm

(20-23)

Talk Title: DataFlow SuperComputing for BigData

Abstract: This presentation analyses the essence of DataFlow SuperComputing, defines its advantages and sheds light on the related programming model. DataFlow computers, compared to ControlFlow computers, offer speedups of 20 to 200 (even 2000 for some applications), power reductions of about 20, and size reductions of also about 20. However, the programming paradigm is different, and has to be mastered. The talk explains the paradigm, using Maxeler as an example, and sheds light on the ongoing research in the field. Examples include: Trading and Finances, CreditDerivatives and numerous related BankingApplications, SignalProcessing, GeoPhysics, WeatherForecast, OilGas, DataEngineering, DataMining, SmartGrid, ScientificSimulations, BrainResearch, Genomics, etc.

Also, a recent study from Tsinghua University in China is presented, which reveals that, for Shallow Water Weather Forecast (a BigData problem), on the 1U level, the Maxeler DataFlow machine is 14 times faster than the Tianhe machine, at the time of the study, rated #1 on the Top 500 list (based on Linpack, which is a smalldata benchmark). Given enough time, the talk also gives a tutorial about the programming in space, which is the programming paradigm used for the Maxeler dataflow machines (established in 2014 by Stanford, Imperial, Tsinghua, and the University of Tokyo). The talk concludes with selected examples and a tool overview (appgallery.maxeler.com and webIDE.maxeler.com). A more detailed tutorial on programming in space will be available after the talk, including the information on MaxGenFD. Related hands-on activities will be performed by remote login (maxeler.mi.sanu.ac.rs). Since December 2016, Maxeler is also available via Amazon AWS. In December 2016, Hitachi of Japan announced its partnership with Maxeler (also available via Amazon AWS), stating that, for their finance and cryptography applications, Maxeler was orders of magnitude faster than any other ControlFlow platform (i.e., CPU or GPU).

 
May 23 Academic partners & IBM Research ALL DAY (Auditorium) Computational Health Summit http://researcher.watson.ibm.com/researcher/view_group_subpage.php?id=7848
May 25

Roger Gould, M.D.

 

Board-certified psychiatrist, psychoanalyst, book author and former Head of Community Psychiatry and Outpatient Psychiatry at UCLA

 

Talk Title: A SECOND CHANCE FOR COGNITIVE COMPUTING AND PSYCHIATRY

Abstract: Back in Boston in the 60’s the AI community was hoping Eliza would grow up to become a digital therapist.  It didn’t work.  Why? That’s the first question along with a number of other “why”s” to be addressed in this talk—Why do so many people intend to change but don’t even when their health is on the line?  Why don’t diabetics take their insulin; or people on a diet stick to it; or people who drink too much cut down? Having an answer for these questions is the core challenge of behavioral health. Yet the cognitive-behavioral health community doesn’t seem to have the right model to answer these questions and solve this problem yet. As a psychiatrist, Dr. Gould has spent thousands of hours listening to patients struggle with these issues, and studied how people think and decide to act or not act. But that is not enough. Extracting that knowledge and making it digital is what is required to be useful. In this talk, Dr. Gould will present his work in Digital Therapy and explain how he thinks psychiatry can help solve the big problem of simply getting people to do what is in their best interest.

Bio: Roger Gould, MD is one of the world's leading physicians on emotional eating and adult development. A board-certified psychiatrist, psychoanalyst, author and former head of Community Psychiatry and Outpatient Psychiatry at UCLA, Dr. Gould has pioneered the use of online therapy sessions focusing on weight loss and stress management in an effort to make the benefits of therapy accessible and affordable to anyone who needs it. The critical thinking process and problem-solving techniques presented in the interactive programs were developed and refined by clinical experience with over 100,000 people and dozens of hospitals and healthcare organizations. The Smithsonian Institution recognized Dr. Gould as a pioneer in the emerging field of computer-assisted therapy. His work has been featured on national television (ABC, CBS and FOX) and covered in Time Magazine, The New York Times, Bariatric Times, Prevention, Psychology Today, Good Housekeeping, MedHelp and MSN Health. He is the author of two books Transformations and Shrink Yourself. 

 

Link to Slides.

 

 

Aug 24 Professor Susan Michie, Professor of Health Psychology and Director of the Center for Behavior Change at University College London 

 Yorktown  Auditorium

9:30AM-11AM

Talk Title: Behavior change intervention design and evaluation: systematic methods for increasing efficiency and effectiveness

Abstract: Behaviour change is an essential ingredient in tackling major health problems such as obesity and cancer. Despite significant investment in programs to change behaviour, interventions to change behaviour vary greatly in their success.  Answering ‘What works, compared to what, how well, for whom, in what settings, for what behaviours and why?’ remains problematic for researchers, policy-makers and practitioners. Efforts to synthesise evidence have traditionally been hindered by poorly reported intervention evaluations, and by the sheer scale on which reports are published: vaster than humans can synthesise and access.

Behaviour change interventions are highly complex, typically comprising several interacting component behaviour change techniques, delivered in different ways and in a variety of contexts. Standardising methods to specify behaviour change interventions, such as their content (the potentially active ingredients within interventions), and to apply theory to intervention development and evaluation in a more rigorous fashion is an active area of research in the behaviour change field where cross-disciplinary research is key to progress.

This talk will present examples from a programme of research aiming to advance the science of behaviour change including the Behaviour Change Wheel and the Human Behaviour Change Project. The latter brings together behavioural and computer scientists with system architects to build an Artificial Intelligence system comprising (i) a Behaviour Change Intervention Ontology, (ii) a machine learning system capable of extracting and interpreting evidence from published literature and making inferences where evidence is lacking, and (iii) an interface allowing users to obtain up-to-date evidence answering their specific questions.

 

 

Sep 6 Pedram Heydari, PhD Univ of California, San Diego and visiting student research collaborator at Princeton Univ 26-014/024 Yorktown

Talk Title: A Theory of Behavior in the Face of Conflicting Inner Motives

Abstract: We propose and axiomatize a multi-attribute stochastic choice model that simultaneously generalizes the classical deterministic choice via utility maximization and the Luce rule. In our model, attributes are cardinal and independent measures of desirability and are endogenously inferred from observed choices. In this model, an item is chosen with positive probability from a set of available items (menu) if and only if it is undominated attribute-wise in that menu. This randomness of choice reflects the possible lack of decisiveness in presence of conflicting inner motives due to the multiplicity of attributes. Also, the model leads to context-dependent evaluation of items by assigning every menu a distinct reference point relative to which the menu items are re-evaluated. This reference point can be interpreted as a context-dependent version of the commonly used status quo or default. Together with the hypothesis of diminishing sensitivity, our model can accommodate a range of empirical findings documenting the effect of context on choices such as the attraction effect, compromise effects, and violations of stochastic transitivity. In addition to these findings, we provide an axiomatic foundation for our choice of reference point for each menu.

Short bio: I am a 5th year Ph.D. student in the Department of Economics at UC San Diego, working with professor Christopher Chambers. My areas of research are in decision theory, behavioral economics, and network economics. In decision theory and behavioral economics, I primarily study situations where there are multiple criteria to evaluate decisions and the patterns that arise as a result, such as randomness of decisions and context-dependence, which can not be accommodated by standard models of choice. In network economics, I study models of network formation and the effect of agents heterogeneity on the structures that arise from an efficiency standpoint as well as that of stability. Since last year, I have been on a visit to Princeton university as a visiting student research collaborator working with professor Faruk Gul.

 

Sep 18
Suchi Saria, Johns Hopkins University

20-001

1-2pm

Talk Title: Machine Learning for Personalised Decision-Making in Healthcare

Abstract: Ever increasing quantities of health-related data are being collected digitally, during clinic visits and through personal monitoring devices. My lab's goal is to enable a new class of diagnostic and treatment planning tools— tools that use statistical machine learning techniques to tease out subtle information from these “messy,” inexpensive observational datasets, and provide reliable inferences for individualizing care decisions. In this talk, I will introduce challenges associated with personalization from observational clinical data. I'll give a preview of a few different examples and then do a deeper dive to discuss one of the results in more detail.

 

Sep 26 Jiaqi Gong 11-12pm 40-000

Talk Title: Enhancing the Real-World Effectiveness of Cognitive Interventions through Smartphones and Wearables

Abstract: Nowadays, users widely accept smartphones and wearables in daily life and even perceive them as part of their self. It provides a window into the relationship between behaviors and mental health. This window is of particular significance to individuals with elevated social anxiety, as it helps to reveal when and where their stress increases in relation to social interactions, ultimately leading to more precise treatment delivery and interventions.

In the collaboration between engineers and psychologists, I will present a three-phase study on using smartphones and wearables for enhancing the real-world effectiveness of cognitive interventions. In phase I, we develop a crowdsourcing platform that integrates smartphones and wearables to understand the relationship between socially anxious individuals' behavior and their anxiety levels. In phase II, we randomly deliver cognitive bias modification interventions to examine the real-world effectiveness. In phase III, we develop just-in-time intervention techniques based on our previous understanding and computational modeling.

I will also present the design considerations and lessons we learned in developing and deploying the crowdsourcing system, discuss the limitations of current work, introduce ongoing work and long-term goals for addressing fundamental challenges of mobile health.

Bio: Dr. Gong is a tenure-track assistant professor in Information Systems at the University of Maryland, Baltimore County. He received his Ph.D. (2010) in Control Science and Engineering from Huazhong University of Science and Technology. His research interests lie in data science and analytics, smart and connected health, and cyber-physical systems. The professional community has recognized his contributions in health research with an NIH scholarship (2015) in mobile health and 7 best paper awards or finalists (2012-2016) at leading international conferences on body sensor networks and wireless health. Also, the more significant endorsement of his research comes from the patients, families, and clinicians with who he had worked to deal with dementia, multiple sclerosis, and other illnesses.



 

Nov 17 Dr. Kun Chen from Uconn department of statistics

11-12pm

20-001

 

Talk Title: Integrate, Divide, and Conquer: On Sparse and Low-Rank Multivariate Statistical Learning --- Kun Chen, Statistics, UConn
Abstract: Large-scale multivariate/multi-view data, or the measuring of distinct yet interrelated sets of characteristics pertaining to a single set of subjects, has become increasingly common. An integrated learning of multiple data attributes, data types or objectives often enables us to gain extraordinary insight of complicated data generation mechanisms through utilizing information from various lenses and angles. To achieve integrative multivariate learning, the general methodology of reduced-rank estimation is one of the most critical ingredients. In this talk, we discuss theory, methodology and computation of several new integrative reduce-rank models, for handling robust dimension reduction, incomplete data, mixed-type data, grouped covariates, among others. In particular, we focus on the challenging problem of recovering a sparse singular value decomposition of the regression component in large-scale multivariate regression. The integrated co-sparse and low-rank structure implies that the outcomes are related to the features through a few distinct pathways, each of which may only involve a subset of outcomes and features. We propose divide-and-conquer strategies to disentangle the multiple coexisting outcome-feature associations. Our approach is able to isolate the recovery of each latent feature, for which the task simplifies greatly to a co-sparse unit-rank regression problem, permitting consistent estimation and scalable computation. The efficacy of the proposed new methods is demonstrated by several applications in genetics, finance and public health studies.
 
 
Speaker's bio: Dr. Kun Chen is an Assistant Professor in the Department of Statistics, University of Connecticut (UConn), and a Research Fellow at the Center for Public Health and Health Policy, UConn Health Center. Chen’s research mainly focuses on integrative multivariate learning, high-dimensional statistics, and healthcare analytics with large-scale heterogeneous data. He has extensive interdisciplinary research experience in a variety of fields including insurance, ecology, biology, agriculture, medical imaging, and public health. Chen’s research projects have received funding from the National Institutes of Health (NIH), the Simons Foundation, and the National Science Foundation (NSF). Recently he is funded by NSF for developing integrative multivariate methods and heterogeneous response regression, and is a co-PI in an NIH-funded data-driven suicide prevention study, which aims to leverage integrated big data from disparate sources in healthcare systems. Chen serves as Secretary of the newly established New England Statistical Society. He is an Associate Editor of Sankhya: The Indian Journal of Statistics since 2016, and has received Recognition for Teaching Excellence at UConn for multiple times.
Chen received his B.Econ. in Finance and Dual B.S. in Computer Science & Technology from the University of Science and Technology of China in 2003, his M.S. in Statistics from the University of Alaska Fairbanks in 2007, and his PhD in Statistics from the University of Iowa in 2011. Before joining UConn, Chen was on the faculty of the Department of Statistics at Kansas State University from 2011 to 2013.
Dec 7 Jiayu Zhou, Assistant Professor in the Department of Computer Science and Engineering at Michigan State University
  • Thursday Dec 7, 2017
  • Time: 11:00 AM - 12:00 PM
  • Location:RCX-801-20-001/NY -
 
Title: Recent Advances of Multi-task Learning
Abstract: The recent decade has witnessed a surging demand in data analysis, where we built machine learning models for various data analysis tasks. The multi-task learning is a machine learning paradigm that bridges related learning tasks and transfers knowledge among the tasks. The seminar reviews multi-task learning basics and recent advances, including distributed multi-task learning that provides efficient and privacy-preserving learning on distributed data sources; and interactive multi-task learning that solicits and integrates domain knowledge in multi-task learning, including human in the learning loop. The seminar is concluded by a discussion of future directions of multi-task learning.
Jiayu-IBM17-small.pdfView Details

 

2016 TALK SCHEDULE
Date Speaker Time & Location Title and additional info Link to Video Streaming, Slides and Replays
Jan. 22  (Friday)  Mudhakar Srivatsa (Watson) 10am (20-001)

Modeling Activities of Daily Living

 

http://researcher.watson.ibm.com/researcher/view.php?person=us-msrivats

Papers:
Demultiplexing Activities of Daily Living in IoT enabled Smarthomes, In InfoCom 2016.
DAISY: A Derivative Inferencing System for Efficient Failure Management in IoT environments, submitted.

Patents:
A system and method for Identifying failed IoT Sensors in Smart Homes using Correlation analysis of multiple sensor streams, filed
A System and Method for Utility based Sensor deployments in Smart Homes for Connected Health and Wellness applications, filed

 

Feb. 25 David Newman Toker (JHU) 11 am (20-001)  

The 5 Worst & 5 Best Ideas for Health Informatics to Improve Diagnostic Safety & Quality

 
March 4 Rong Chen (Mt Sinai) 11 am (20-043) Using big data to interpret genomes for diagnostics, therapeutics, and precision medicine http://rongchenlab.org/people/
March 23

Artemis Simopoulos, MD, FACN

Founder and President, Center for Genetics, Nutrition and Health, Washington, DC

1pm (Auditorium) The omega-6/omega-3 ratio in the prevention and management of Obesity http://americannutritionassociation.org/artemis-simopoulos
March 31 Judith Klein-Seetharaman (University of Warwick) 11am (20-001) Molecular Motivators to Lifestyle Changes   
April 13 Chunhua Weng (Columbia University) 10am (20-001) Optimizing the Design and Conduct of Clinical Research with Informatics. http://people.dbmi.columbia.edu/~chw7007/
April 18 Hold for WH seminar 1-2:30pm (Auditorium) HealthKit and ResearchKit, the Evolution of the Apple Health Ecosystem  
April 21 Ty Ridenour PhD MPE, Developmental Behavioral Epidemiologist, Behavior and Urban Health Program, RTI 1pm (20-043)
Rigorous Within-person Analytics: Illustrations and Applications
 
May 9 Prof. Miguel Hernan from the Biostatistics Department at Harvard School of Public Health 11am (20-001)
Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available
 
May 19   Corey Arnold (Assistant Professor in the Medical Imaging and Informatics group at the University of California - Los Angeles) 10:30-12pm (ThinkLab Theater)   Exploring Unstructured Biomedical Data using Topic Models  
 May 19  
Heng Luo (NuMedii, Inc.)
 1-2pm (03-235)  
Chemical-protein interactome and its application in drug development
 
May 20 Hanghang Tong (Assistant Professor, Arizona State University) 1-2pm (20-059) Inside the Atoms: Mining a Network of Networks and Beyond  
June 24

Prof. Riccardo Bellazzi

(University of Pavia, Italy)

10-11am (20-001) the implementation and the results obtained in two European projects, Mosaic and Mobiguide, dealing with Type 2 Diabetes and Atrial Fibrillation, respectively, and on the system implemented to support clinical research in Oncology at the Fondazione S. Maugeri hospital of Pavia.  
July 7 Mark L Graber 2-3pm 20-001
Diagnostic Error and Health IT
 

2015 Schedule:

Date Speaker Time & Location Title and additional info Link to Video Streaming, Slides and Replays
Apr 30

Patrick Ryan

 

11am-noon

YKT 20-043

An Open Science Approach to Medical Evidence Generation:  Introducing Observational Health Data Sciences and Informatics



Sr. Director and Head, Epidemiology Analytics

Janssen Research and Development

 

http://www.ohdsi.org/who-we-are/collaborators/1045-2/

 

 

Slides for April 30 meeting

Link to video streaming:

http://watkvideo01.watson.ibm.com/Mediasite/Play/caf5e1b010ff4a1695666bc40c0a71151d

Link to the replay of video streaming:

http://watkvideo01.watson.ibm.com/Mediasite/Play/baf9d067eed54802924887f677b54aa41d

May 15 Noemie Elhadad

5.15 10:30

confirmed 5.6.15

Natural Language Processing to Support Clinicians at the Point of Care

Chair of Health Analytics Center, Data Science Institute

Associate Professor, the Department of Biomedical Informatics

Columbia University

 

http://people.dbmi.columbia.edu/noemie/

Noemie-15ibm.pdf

 

Link to video streaming:

http://watkvideo01.watson.ibm.com/Mediasite/Play/80c07755ec7d418ca6bd0a213e1f8bc91d

May 20 Jack Gallant

5.20

11am

Auditorium

(as part of the Brain Day event)

 

Reverse-engineering the human brain with neuroimaging and big data

Jack L. Gallant, UC Berkeley

 

http://www.gallantlab.org/

http://www.nature.com/news/brain-decoding-reading-minds-1.13989

Co-hosted with Multimedia PIC

 

The livestream link will be here:



https://bluestream.somers.ibm.com/v3/research/

May 27 Veljko Milutinovic 5.27

DataFlow SuperComputing for BigData Analytics

 

Prof. Veljko Milutinovic, University of Belgrade, IEEE Fellow

DataFlow SuperComputing for BigData Analytics

Co-hosted with SuperComputing PIC
May 29 Zhaonan Sun 5.29

Multi-task Learning approach for Comprehensive Risk Prediction

Zhaonan Sun, Healthcare Analytics Group

IBM

Reference:

Zhaonan Sun, Fei Wang, Jianying Hu. LINKAGE: An Approach for Comprehensive Risk Prediction for Care Management. KDD 2015 (To Appear)

June 5 Werner Doyle

 

2:30pm-3:30pm

Auditorium

 

Neurosurgery: A Tool for Understanding and Altering the Human Nervous System to Sustain its Well being

Professor, NYU Langone Medical Center Neurosurgery Dept.

Neuromodulation for comprehensive epilepsy

http://spectrum.ieee.org/biomedical/bionics/life-as-a-bionic-woman

http://nyulangone.org/locations/comprehensive-epilepsy-center

 

ThinkFriday event

Co-hosted with Physics Science Seminar Series 

June 8 Harold Lehmann 12-1 20-001

The use of EHR data for clinical research: Experience of the PaTH Clinical Data Research Network

Professor and Interim Director, Division of Health Sciences Informatics

Johns Hopkins School of Medicine​

Link to video recording (apologies- for an unknown reason, only the first 15 minutes were recorded): 

http://watkvideo01.watson.ibm.com/Mediasite/Play/ef018affe883425095e7409d85acdafa1d

June 11 Sufi Zafar  

Dr. Sufi Zafar on Biosensors

 

 
July 28 Sally Okun

1:00-2:30pm 

20-001

Patient-like-me 

Speaker: Sally Okun, RN, MMHS

PatientsLikeMe: Harnessing the Power of Patient Voice

Established in 2004, PLM has been an early pioneer as a patient-powered research network. Today, over 350,000 members have reported their real-world experiences on more than 2,500 diseases, ranging from rare conditions such as amyotrophic lateral sclerosis (ALS) to more prevalent ones like depression, fibromyalgia, multiple sclerosis, and psoriasis.  Members create health profiles using clinically relevant and research-based data collection tools to monitor how they’re doing between doctor or hospital visits, document the severity of their symptoms, identify triggers, note how they are responding to new treatments, and track side effects.  Members connect with each other in various ways including Forum discussions, journal feeds and private messaging.

     This presentation will explore how patient voice, the missing link in health data, is driving a new era of discovery, research and care – one in which people can benefit in real time from the information they share while contributing to a new way of generating, measuring and aggregating health insights that can benefit each of us. 

Sept? Joel Saltz  

Head of Biomedical Informatics Department

Stony Brook University

 

http://sb.cc.stonybrook.edu/happenings/homespotlight/dr-joel-saltz-to-head-new-biomedical-informatics-department/

 

 
Aug (First week) Professor John Rogers, UIUC Materials Science and Engineering   miniaturized flexible EEG and harvesting energy from the body  
Sep Mark Wightman  

Neuroscience and Analytical Chemistry Group

University of North Carolina at Chapel Hill

 

Microelectrodes and their use to probe complex chemical and biochemical phenomena

http://www.chem.unc.edu/people/faculty/wightman/group/

 

The Wightman Group develops and utilizes fast-scan cyclic voltammetric (FSCV) methods to monitor sub-second neurotransmission events in the brain. Very small microelectrodes coupled to rapid electrochemical monitoring enable spatially- and temporally-resolved measurements of phasic neurotransmitter release in tissue slice preparations, anaesthetized animals, and freely-moving, behaving animals.

 

Co-hosted with Physics Science Seminar Series 
Sep 30 Eric Karl Oermann, M.D.  

Title: Machine learning with medical data to predict future clinical states

Short Bio: Eric Karl Oermann is a houseofficer in the Department of Neurological Surgery at the Mount Sinai Health System. He received his undergraduate degree in mathematics from Georgetown University with a concentration in differential geometry. After spending time with the President's Council on Bioethics studying human dignity, he attended medical school at Georgetown University from 2008 to 2013. While a medical student, he won a grant from the American Brain Tumor Association (ABTA) to study the epigenetics of the IDH1 gene utilizing somatic cell gene targeting, and received a Doris Duke Foundation Fellowship to spend a year at UNC Chapel Hill studying glioma epigenetics and artificial neural networks. While a houseofficer, he was selected by Forbes Magazine as one of the 30 Under 30 in Healthcare. He has published over 30 peer reviewed articles on clinical neurosurgery, cancer epigenetics, machine learning, and bioethics. His clinical interests include stereotactic radiosurgery, functional neurosurgery, and tumors of the central nervous system, and his research interests include machine learning, deep learning, and mind-machine interfaces.

Host: Ping Chang (invited by Gustavo
Oct 26

Isaac Kohane

1-2 pm

YKT 20-001

also booked auditorium

Harvard Medical School

Henderson Professor of Pediatrics and Health Sciences and Technology
Co-Director Center for Biomedical Informatics Harvard Medical School
Director of the Francis A. Countway Library of Medicine

 

https://cbmi.med.harvard.edu/people/isaac-kohane

 

Dr. Kohane leads multiple collaborations at HMS and is hospital affiliates in the use of genomics and computer science to study cancer and the development of the brain (with emphasis on autism). He has developed several computer systems to allow multiple hospital systems to be used as “living laboratories” to study the genetic basis of disease while preserving patient privacy. Dr. Kohane’s research builds on his doctoral work in computer science on decision-support and subsequent research in machine-learning applied to biomedicine. Dr. Kohane leads several NIH-funded efforts to translate genomic research into clinical practice and continues his own practice in pediatric endocrinology at Boston Children’s Hospital.

 

 
Oct Dubois Bowman  

Chair the Mailman School of Public Health, Dept. of Biostatistics

Professor of Biostatistics

http://www.mailman.columbia.edu/our-faculty/profile?uni=fb2403

 

 
  David Newman Toker

YKT 20-001

1-2

 

email from Vanessa Liewellyn 7.24 saying he can't make it is Oct.

rescheduled for 2.25.16 (email from Vanessa 10.19.15)

Nov 20 John Greally

YKT 20-001

10:45-12:00

Albert Einstein