Center for Predictive Computational Phenotyping (CPCP): Training Plans

May 15, 2015

Debora Treu and Whitney SweeneyCenter for Predictive Computational Phenotyping (CPCP)

University of Wisconsin-Madison

What is CPCP?• The Center for Predictive Computational Phenotyping

(CPCP) promises to significantly advance the state of the art in computational methods for transforming large, heterogeneous, high-dimensional data sources into predictive models for biomedicine. Specifically, we are focusing on a broad range of problems that can be cast as computational phenotyping.

timenow

predict a clinically important phenotype before it is exhibited

prospective phenotyping

extract a relevant phenotype from a complex data source or collection of sources

retrospective phenotyping

?

The term phenotype refers to the observable properties of an organism that result from the interaction of its genotype and its environment. Some phenotypes are easily measured and interpreted, and are available in an accessible format. However, a wide range of scientifically and clinically important phenotypes do not satisfy these criteria. In such cases, computational phenotyping methods are required either to:

CPCP Aims• Aim 1

Develop, evaluate and disseminate computational and statistical algorithms, models and software packages that significantly advance the state of the art in predictive computational phenotyping.

• Aim 2

Develop, conduct and evaluate training activities that reach a broad set of audiences whose education, research and practice would significantly benefit from having state-of-the-art knowledge about data science, predictive models for biomedicine, and computational phenotyping. These audiences include:

• Biomedical scientists• Clinicians• Data scientists• Postdocs• Graduate students• Undergraduates• General public

biomedical informatics

computer science

biostatistics

engineering

biological sciences

bioethics & law

medicine

CPCP

The Center for Predictive Computational Phenotyping

Types of Events

• Weekly Center Meetings• Short courses• Tutorial Modules• “Boot camp” Tutorials• Seminars• Training in Undergraduate Research• TED-style talks• Annual Center Retreat

Summer Research Program in Computational Biology and Biostatistics (CBB)

• Nicholas Escanilla, Lake Forest College Mentor: David Page

• Vir Patel, Western Kentucky University Mentor: Sushmita Roy

• Emily Balczewski, Carleton College Mentor: Vikas Singh

Seminars on High-Throughput Computing Platforms

• The primary goal of these seminars are to help researchers make sense of what tools are

available for the analysis of large and complex data sets and how they are relevant to their

specific work.

• The Hadoop Ecosysytem: A Bird's-Eye Overview

Jignesh Patel, CS

This seminar addressed tools like Hadoop

and Hive. It was videotaped and will be

available on the CPCP website.

• CONDOR and the Center for High Throughput Computing Miron Livny, CS and Morgridge; Tim Cartwright, CS; Lauren Michael & Christina Koch, Morgridge

This seminar will specifically deal with the tools provided by the Center

for High Throughput Computing at the UW Madison. This seminar is tentatively

scheduled for June 2015. It will also be videotaped and available on the CPCP website.

Small Talks on Big Data• These “TED-style” talks have been created to introduce

some of the scientific questions and quantitative problems with which the CPCP is grappling. Our goals are to share the knowledge and innovations derived while working on these issues as well as to promote collaborations with other data and biomedical scientists.• Mark Craven, BMI

• Biomedical Big Data and the Computational Phenotyping Challenge

• Beth Burnside, Radiology• Personalizing Breast Cancer: Integrating Predictive Phenotypes into

Clinical Care

• David Page, BMI• Predicting Health Events from Electronic Health Records by Machine

Learning

First Annual Retreat

Center for Predictive Computational Phenotyping June 12, 2015

DeLuca Forum, The Discovery Building

8:00 – 8:30 continental breakfast and registration

8:30 – 8:45 Overview of the Center for Predictive Computational Phenotyping

Mark Craven, Professor, Dept. of Biostatistics & Medical Informatics (BMI)

Director, CPCP

8:45 – 9:00 welcoming remarks

Robert Golden, Dean of the School of Medicine and Public Health

9:00 – 10:15 keynote talk

Understanding Alzheimer’s via Neuroimage-Based Phenotyping

Barbara Bendlin, Assistant Professor, Dept. of Medicine

Sterling Johnson, Professor, Dept. of Medicine

Vikas Singh, Associate Professor, Dept. of BMI

10:15 – 11:00 poster session

11:00 – 12:00 Transcription-Based Cellular Phenotyping

Colin Dewey, Assoc. Professor, Dept. of BMI

Christina Kendziorski, Professor, Dept. of BMI

Epigenome-Based Phenotyping

Sunduz Keles, Professor, Dept. of BMI

Sushmita Roy, Assistant Professor, BMI

12:00 – 1:00 lunch

(continued)

1:00 – 1:15 welcoming remarks

Brad Schwartz, CEO of the Morgridge Institute for Research

1:15 – 2:45 Prioritization Strategies for Ranking, Selection and Data Integration in

Computational Phenotyping

Michael Newton, Professor, Dept. of BMI

Associate Director, CPCP

High-Throughput Computing in Support of High-Throughput Phenotyping

Miron Livny, Professor, Dept. of Computer Sciences

CTO, Morgridge Institute

Gary Pack, Postdoctoral Fellow, Marshfield Clinic Research Foundation

Yuriy Sverchkov, Postdoctoral Fellow, Dept. of BMI

Scaling up Inductive Logic Programming for EHR-Based Phenotyping

David Page, Professor, Dept. of BMI

Jignesh Patel, Professor, Dept. of Computer Sciences

2:45 – 3:30 poster session

3:30– 4:30 Active Learning from Human Experts and an Application to

EHR-Based Phenotyping

Rob Nowak, Professor, Dept. of Electrical and Computer Engineering

Peggy Peissig, Director, Biomedical Informatics Research Center

Marshfield Clinic Research Foundation

Building Entity Matching Management Systems for Big Data Problems in

Biomedicine

AnHai Doan, Professor, Dept. of Computer Sciences

4:30 – 4:45 closing comments

4:45 – reception

The CPCP partners with the Morgridge Institute for Research and the Marshfield Research Clinic, and receives additional support from the UW-Madison School of Medicine and Public Health and the Office of the Vice Chancellor for Research and Graduate Education at the University of Wisconsin-Madison with funding from the Wisconsin Alumni Research Foundation.

This research is supported by the National Institutes of Health under Award Number U54AI117924. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.