I blend data visualization and data mining techniques to create visual interactive systems to help users make sense out of big data. Lately, my research focuses on extracting insights from clinical data to support data-driven medicine. -@adamperer
Adam Perer is currently based in Pittsburgh, where he is a Research Scientist at IBM's T.J. Watson Research Center, where he is a member of the Healthcare Analytics Research Group. He is also an Adjunct Professor in the Human-Computer Interaction Institute at Carnegie Mellon University. His research in visualization and human-computer interaction focuses on the design of novel visual analytics systems. This work has been published at premier venues in visualization, human-computer interaction, and medical informatics (IEEE InfoVis, IEEE VAST, ACM CHI, ACM CSCW, ACM IUI, AMIA). He holds Ph.D. and masters degrees in Computer Science from the University of Maryland, College Park, and a B.S. in Computer Science from Case Western Reserve University.
Clustering, the process of grouping together similar items into distinct partitions, is a common type of unsupervised machine learning that can be useful for summarizing and aggregating complex multi-dimensional data. However, data can be clustered in many ways, and there exist a large body of algorithms designed to reveal different patterns. While having access to a wide variety of algorithms is helpful, in practice, it is quite difficult for data scientists to choose and parameterize algorithms to get the clustering results relevant for their dataset and analytical tasks. To alleviate this problem, we built Clustervision, a visual analytics tool that helps ensure data scientists find the right clustering among the large amount of techniques and parameters available.PAPER Bum Chul Kwon, Ben Eysenbach, Janu Verma, Kenney Ng, Christopher deFilippi, Walter F. Stewart, and Adam Perer. Clustervision: Visual Supervision of Unsupervised Clustering. IEEE Visual Analytics Science and Technology (VAST). Phoenix, Arizona, USA. (2017).
Understanding predictive models, in terms of interpreting and identifying actionable insights, is a challenging task. Often the importance of a feature in a model is only a rough estimate condensed into one number. However, our research goes beyond these naive estimates through the design and implementation of an interactive visual analytics system, Prospector. By providing interactive partial dependence diagnostics, data scientists can understand how features affect the prediction overall. In addition, our support for localized inspection allows data scientists to understand how and why specific datapoints are predicted as they are, as well as support for tweaking feature values and seeing how the prediction responds.PAPER Josua Krause, Adam Perer, and Kenney Ng. Interacting with Predictions: Visual Inspection of Black-box Machine Learning Models. ACM Conference on Human Factors in Computing Systems (CHI 2016). San Jose, California. (2016).
Many researchers aim to analyze the behavior of cohorts whose behaviors are recorded in large event databases. However, extracting cohorts from databases is a difficult step, often overlooked. This is especially true when researchers wish to restrict their cohorts to exhibit a particular temporal pattern of interest. COQUITO is a visual interface that assists users defining cohorts with temporal constraints, designed to be comprehensible to domain experts with no preknowledge of database queries and also to encourage exploration.PAPER Josua Krause, Adam Perer, and Harry Stavropolous. Supporting Iterative Cohort Construction with Visual Temporal Queries. IEEE Visual Analytics Science and Technology (VAST). Chicago, USA. (2015).
Predictive modeling techniques are increasingly being used by data scientists to understand the probability of predicted outcomes. However, for data that is high-dimensional, a critical step in predictive modeling is determining which features should be included in the models. INFUSE is a novel visual analytics system designed to help analysts understand how predictive features are being ranked across feature selection algorithms, cross-validation folds, and classifiers.PAPER Josua Krause, Adam Perer, and Enrico Bertini. INFUSE: Interactive Feature Selection for Predictive Modelling of High Dimensional Data. IEEE Visual Analytics Science and Technology (VAST). Paris, France. (2014).
Frequence and Care Pathway Explorer are visual analytics system that integrates data mining and visualization for finding frequent patterns from longitudinal event sequences. They features novel frequent sequence mining algorithms to handle multiple levels-of-detail, temporal context, concurrency, and outcome analysis. Additionally, they also feature visual interfaces designed to support insights, and support exploration of patterns of the level-of-detail relevant to users.PAPER Adam Perer and Fei Wang. Frequence: Interactive Mining and Visualization of Temporal Frequent Event Sequences. ACM Intelligent User Interfaces (IUI) 2014. Haifa, Israel. (2014).
As datasets grow and analytic algorithms become more complex, the typical workflow of analysts launching an analytic, waiting for it to complete, inspecting the results, and then re-launching the computation with adjusted parameters is not realistic for many real-world tasks. We present an alternative workflow, progressive visual analytics, which enable analysts to inspect partial results of an algorithm as they become available and interact with the algorithm to prioritize subspaces of interest.PAPER Charles Stolper, Adam Perer, and David Gotz. Progressive Visual Analytics: User-Driven Visual Exploration of In-Progress Analytics. IEEE Visual Analytics Science and Technology (VAST). Paris, France. (2014).
Cohort analysis is a widely used technique for the investigation of risk factors among clinical populations. However, this process is often constrained by a lack of integrated analytics and visualization. To address this challenge, we have designed CAVA, a platform for Cohort Analysis via Visual Analytics, which is designed to help domain experts work faster and more independently when performing retrospective cohort studies.PAPER Zhiyuan Zhang, David Gotz, and Adam Perer. Iterative Cohort Analysis and Exploration. Journal of Information Visualization (IVS). (2014).
Common chronic diseases such as hypertension are costly and difficult to manage, so we explore using predictive modeling and visualization to predict the risk and timing of deterioration in hypertension control.PAPER Jimeng Sun, Candace D McNaughton, Ping Zhang, Adam Perer, Aris Gkoulalas-Divanis, Joshua C Denny, Jacqueline Kirby, Thomas Lasko, Alexander Saip, Bradley A Malin Predicting changes in hypertension control using electronic health records from a chronic disease management program. Journal of the American Medical Informatics Association (JAMIA). 21(2): 337-344 (2014).
CareFlow is a novel visual analytics tool designed to help clinicians devise care plans for their patient. Using historical outcomes from clinically similar patients, CareFlow allows doctors to analyze which treatments have been effective for patients like theirs.PAPER Adam Perer and David Gotz. Data-Driven Exploration of Care Plans for Patients. ACM CHI 2013. Paris, France. (2013).
Patients' medical conditions often evolve in complex and seemingly unpredictable ways. Even within well-defined episodes of care, variations between patients' progression and outcome can be dramatic. We present a visual analytic system to support outcome analysis from retrospective clinical patient data. Our approach combines (1) visual queries to specify episodes, (2) pattern mining techniques to discover important events, and (3) interactive visualizations uncover event patterns that impact outcome and change over time.PAPER David Gotz, Fei Wang, and Adam Perer. A Methodology for Interactive Mining and Visual Analysis of Clinical Event Patterns using Electronic Health Record Data. Journal of Biomedical Informatics (JBI). (2014).
MatrixFlow is a visual analytic tool designed to help aid clinical researchers by making the subtle trends of disease progression more obvious. By unearthing the hidden patterns of co-occuring symptoms in patient health records, emerging health risks may become more discoverable.PAPER Adam Perer and Jimeng Sun. MatrixFlow: Temporal Network Visual Analytics to Track Symptom Evolution during Disease Progression. American Medical Informatics Association Annual Symposium (AMIA 2012). Chicago, Illinois. (2012).
Orion is a system for interactive modeling, transformation and visualization of network data. Orion’s interface enables the rapid manipulation of large graphs — including the specification of complex linking relationships — using simple drag-and-drop operations.PAPER Jeffrey Heer and Adam Perer A System for Modeling, Transformation and Visualization of Multi-dimensional Heterogeneous Networks. Information Visualization Journal (2014).
SaNDVis is a visual analytics tool that mines, aggregates, and infers a social graph from social media. The visualization tool supports people-centric tasks like expertise location, team building, and team coordination in the enterprise.PAPER Adam Perer, Ido Guy, Erel Uziel, Inbal Ronen, Michal Jacovi. The Longitudinal Use of SaNDVis: Visual Social Network Analytics in the Enterprise. IEEE Transactions in Visualization and Computer Graphics (IEEE TVCG). (2013).
While working in the Social Technologies Group at IBM Research Haifa, my research centered around methods and techniques for making sense of big data in social media.PAPER Michael Muller, Kate Ehrlich, Tara Matthews, Adam Perer, Inbal Ronen, and Ido Guy. Diversity among Enterprise Online Communities: Collaborating, Teaming, and Innovating through Social Media. ACM Conference on Human Factors in Computing Systems (CHI 2012). Austin, Texas. (2012).
A large amount of my Ph.D. research at the University of Maryland involved making sense out of social networks. Much of this research was integrated into SocialAction, a novel visual social network analytics tool. A variety of publications were published on this work, including:BOOK CHAPTER Adam Perer. Finding Beautiful Insights in the Chaos of Social Network Visualizations. Beautiful Visualization. O’Reilly Press. (2010).
In response to demand for analyzing massive networks, we developed a new methodology for interacting with networks: "Search, show context, expand-on-demand” where users search for particular datapoints as a focus for analysis and the system then computes and displays an optimal relevant context given the users’ current interests.PAPER Frank van Ham and Adam Perer. “Search, Show Context, Expand on Demand”: Supporting Large Graph Exploration with Degree-of-Interest. IEEE Conference on Information Visualization (InfoVis). Atlantic City, New Jersey, USA. (2009).
Email, while often considered personal and private, is also a rich source of social data. Over time, people accumulate extensive email repositories that contain detailed information about their personal communication patterns and relationships. My prior research focused on building visualizations that capture hierarchical, correlational, and temporal patterns present in email repositories.PAPER Adam Perer and Marc A. Smith. Contrasting portraits of email practices: visual approaches to reflection and analysis. International Conference on Advanced Visual Interfaces (AVI). Venice, Italy. (2006).