Recent & Upcoming Events

Nov 22, 2016

Big Data in Behavioral Medicine

Understanding the genesis of and developing interventions for adverse health-related behaviors

Oct 18, 2016

Big Privacy: Policy Meets Data Science Symposium

This symposium examined the legal, policy, and technical issues at the intersection of data privacy

Jun 30, 2016

CPCP Second Annual Retreat

A day-long retreat highlighting recent research in the CPCP

Apr 21, 2016

CPCP Seminar: Mining Structures from Massive Bio-Text Data: A Data-Driven Approach by Dr. Jiawei Han

Jiawei Han from the BD2K KnowEng Center-UIUC discussed mining structures from massive bio-text data

Nov 10, 2015

CPCP Seminar: Transforming Your Research with High-Throughput Computing by Lauren Michael

Lauren Michael from the CHTC discussed high-throughput computing approaches to Big Data

Training Resources

CPCP Privacy Symposium 2016: Privacy Preserving Federated Biomedical Data Analysis Symposium Video

Learn about the challenges associated with the technical approaches for utilizing data from multiple sources to build more accurate machine learning algorithms from Dr. Xiaoquian Jiang. We know that having more types of data and data from distributed sources provides a stronger platform for research and discovering with machine learning. To address privacy in this context, Dr. Jiang proposes a privacy-preserving distributed data framework and describes various models implemented to solve the such problems. Dr. Jiang's research group has produced versions of this framework in R and Java as well as an online web-service. All version of this framework are available for other researchers to use for their own analysis.

CPCP Privacy Symposium 2016: Privacy is an Essentially Contested Concept Symposium Video

What does privacy mean in the context of Big Data? Dr. Deirdre Mulligan discusses various definitions of privacy in law, philosophy and computer science. Traditional approaches to privacy in data place most of the responsibility for the control of private information flow on the individual with mechanisms such as consent. This idea, known as informational actualization, has limitations that have been exposed by machine learning on big data. These limitations cause violations of privacy such as uncovering identity of individuals where it has been withheld or unexpected inferences made from data that have been intentionally disclosed. Dr. Mulligan suggests new ways of viewing privacy that evolve as social life and technology change.

CPCP Privacy Symposium 2016: Proving that Programs Do Not Discriminate Symposium Video

As the field of Artificial Intelligence (AI) continues to advance, an increasing number of prediction are made by computer programs about humans. These predictions affect decisions made about humans in a wide variety for areas including decisions about: who should get the job, the bank loan, or early release from prison. As we increasingly rely on AI programs to help make decisions about peoples lives, it becomes vitally important that we are able to ensure the programs we are depending on do not have an unfairly biased against certain groups of people. Dr. Aws Albarghouthi of the University of Wisconsin - Madison Computer Sciences department uses his expertise in programming languages to address this issue of fairness.

CPCP Privacy Symposium 2016: Panel Discussion Symposium Video

Dr. Pilar Ossorio from Morgridge Institute for Research at the University of Wisconsin-Madison, and Dr. Peggy Peissig from the Biomedical Informatics Research Foundation join Dr. Aws Albarghouthi, Dr. Deirdre Mulligan, and Dr. Xiaoquian Jiang to answer questions from the audience about privacy and fairness in the context of computational analysis.

CPCP Privacy Symposium 2016: Welcome Symposium Video

Dr. Pilar Ossorio and Dr. Mark Craven welcome attendees to the 2016 Big Privacy Symposium. At the Center for Predictive Computational Phenotyping (CPCP) we have expertise in the ethical and legal aspects of analyzing complex datasets. Experts in these fields work with our experts in computational analysis to ensure that new methods for improving human health developed at CPCP are discovered and used in a legally and ethically manner.

Recent Publications

Structure-leveraged methods in breast cancer risk prediction. Fan J, Wu Y, Yuan M, Page D, Liu J, Ong IM, Peissig P, Burnside E. Journal of Machine Learning Research 17:1-15, 2016

Modeling the temporal evolution of postoperative complications. Feld SI, Cobian AG, Tevis SE, Kennedy GD, Craven MW. Proceedings of the American Medical Informatics Association Annual Symposium, 2016

Adaptive signal recovery on graphs via harmonic analysis for experimental design in neuroimaging. Kim WH, Hwang SJ, Adluru N, Johnson SC, Singh V. Proceedings of the 14th European Conference on Computer Vision (ECCV), Volume 9910 Lecture Notes in Computer Science, 2016

Magellan: toward building entity matching management systems. Konda P, Das S, Suganthan P, Doan A, Ardalan A, Ballard JR, Li H, Panahi F, Zhang H, Naughton J, Prasad S, Krishnan G, Deep R, Raghavendra V. Proceedings of the 42nd International Conference on Very Large Databases, 2016

Baseline regularization for computational drug repositioning with longitudinal observational data. Kuang Z, Thomson J, Caldwell M, Peissig P, Stewart R, Page D. Proceedings of the 25th International Joint Conference on Artificial Intelligence, 2016

Recent Resources

MetaSRA: normalized metadata for the Sequence Read Archive data

MetaSRA is an annotation/re-coding of sample-specific metadata in the Sequence Read Archive using biomedical ontologies. Currently, MetaSRA maps biological samples to biologically relevent terms in the Disease Ontology, Experimental Factor Ontology, Cell Ontology, Uberon, and Cellosaurus.

MetaSRA pipeline software

This pipeline was used to construct the MetaSRA database which provides normalized metadata for the Sequence Read Archive .This pipeline re-annotates key-value descriptions of biological samples using biomedical ontologies.

Magellan software

Magellan is a set of tools and how-to guides for developing entity-matching systems. The Magellan tools are built on the Python data science and big data eco-system, and aim to cover the entire entity-matching pipeline.

EBSeqHMM software

EBSeqHMM is an R package that implements an auto-regressive hidden Markov model for identifying genes and isoforms that have expression changes in ordered RNA-seq experiments, and clustering the identified genes into paths showing similar changes. EBSeqHMM is suitable for any ordered RNA-seq experiment including time courses and spatially ordered experiments.

Oscope software

Oscope is a statistical pipeline for identifying oscillatory genes and characterizing one cycle of their oscillation, referred to as a base-cycle, in unsynchronized snapshot single cell RNA-seq experiments. The Oscope pipeline includes three modules: a paired-sine model module to identify candidate oscillator pairs; a clustering module to cluster candidate oscillators into groups; and an extended nearest insertion module to estimate the base-cycle oscillation within each group.