Our paper was presented at AMIA 2012 in Chicago in November 2012.
Citation:
Adam Perer, 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).
Objective: To develop a visual analytic system to help medical professionals improve disease diagnosis by providing insights for understanding disease progression.
Methods: We develop MatrixFlow, a visual analytic system that takes clinical event sequences of patients as input, constructs time-evolving networks and visualizes them as a temporal flow of matrices. MatrixFlow provides several interactive features for analysis: 1) one can sort the events based on the similarity in order to accentuate underlying cluster patterns among those events; 2) one can compare co-occurrence events over time and across cohorts through additional line graph visualization.
Results: MatrixFlow is applied to visualize heart failure (HF) symptom events extracted from a large cohort of HF cases and controls (n=50,625), which allows medical experts to reach insights involving temporal patterns and clusters of interest, and compare cohorts in novel ways that may lead to improved disease diagnoses.
Conclusions: MatrixFlow is an interactive visual analytic system that allows users to quickly discover patterns in clinical event sequences. By unearthing the patterns hidden within and displaying them to medical experts, users become empowered to make decisions influenced by historical patterns.
MatrixFlow: Temporal Network Visual Analytics to Track Symptom Evolution during Disease Progression
1. MatrixFlow
Temporal Network
Visual Analytics
to Track Symptom Evolution
during Disease Progression
Adam Perer
Jimeng Sun
Healthcare Analytics Research Group
IBM T.J. Watson Research Center
2. Overview
•Visual Analytics for Data-Driven
Healthcare
Text Mining
• To extract symptoms from clinical notes
Social Network Analysis
• To model co-occurence symptom networks
Visualization
• To enable clinicians to reach insights
3. Challenge
•Difficult to Diagnose Diseases
•Co-morbidities may mask presence.
•Physicians often use clinical
knowledge without quantitative
data from Electronic Medical
Records.
•There are few analytical tools to
extract meaningful insights.
4. Heart Failure
•Potentially fatal disease that affects
2% of adults in developed countries
•Difficult to manage
•No systematic diagnostic criteria
•Framingham HF Diagnosis based on
•2 major criteria
•1 major criteria & 2 minor criteria
Heart designed by Catherine Please from The Noun Project
6. Population
•50,625 Patients (Geisinger Clinic PCPs)
•4,644 case patients
• 1,200 with preserved ejection fraction
• 1,615 with reduced ejection fraction
• 45,981 control patients
7. Text Mining
•3.3 million Clinical Notes (4 GB of Text)
•NLP used to identify Framingham
screenshot of part of a clinical note.
criteria
•900,000 positive
•3.6 million negative
•97% of HF cases met
criteria extracted
•8% of controls met Jimeng Sun, PhD
criteria extracted
8. Text Mining
•3.3 million Clinical Notes (4 GB of Text)
•NLP used to identify Framingham
screenshot of part of a clinical note.
criteria
•900,000 positive
•3.6 million negative
•97% of HF cases met
criteria extracted
•8% of controls met Jimeng Sun, PhD
criteria extracted
9. Sequence
HF Diagnosis
Ankle Edema
Medication 1
DO Exertion
Medication 2
DO Exertion
AP Edema
Medication 3
10. Sequence
HF Diagnosis
DO Exertion
Ankle Edema DO Exertion AP Edema
Medication 1 Medication 2 Medication 3
Year 1 Year 2 Year 3
11. Network
HF Diagnosis
DO Exertion
Ankle Edema DO Exertion AP Edema
Medication 1 Medication 2 Medication 3
Year 1 Year 2 Year 3
12. Network
Ankle Edema
HF Diagnosis
Medication 1
DO Exertion
DO Exertion AP Edema
Medication 2 Medication 3
Year 1 Year 2 Year 3
13. Network
Ankle Edema DO Exertion
HF Diagnosis
Medication 1 Medication 2
DO Exertion
AP Edema
Medication 3
Year 1 Year 2 Year 3
14. Network
Ankle Edema DO Exertion DO Exertion AP Edema
Medication 1 Medication 2 Medication 3
Year 1 Year 2 Year 3
16. xperienced Symptom1 and Symptom 3 in the same time interval, then there would not be an edge
Clustering
those two events. This co-occurrence network is computed using our advanced network modeling
k, Orion10. As our networks now feature edges that have varying edge weights, our matrix visualiza
e color of each cell according to a sequential color scale representing the edge weight value (such as
e shown at the bottom of Figure 6).
•Visualization reveals clusters of
g Clinical Event Networks
clinical events
ualizing matrices, it is important to choose an effective method to sort the order of nodes in order to
atterns as possible4. As we wish to reveal clusters of clinical events, we employ a greedy hierarchica
optimizing Newman’s modularity metric17. From this algorithm, we are able to obtain a sort order t
•Hierarchical Clustering (Newman’s
the distance among connected nodes by ordering the nodes according to the cluster tree produced b
al clustering algorithm. Figure 4 shows an example of matrix visualization before and after the sorti
modularity matrix)
In the latter example, well-connected nodes (APEdema, DOExertion, and Medication 3) assemble
ge box-like structure in the visualization, which makes the cluster more apparent than in the unsorte
•Determines sort order for matrices
the left.
17. Population
•Align by diagnosis date
•Aggregate patient event networks
•Split by user-chosen intervals
Patient 1 A B C Diagnosis
Patient 2 B C E Diagnosis
Patient 3 A B D Diagnosis
Patient 4 C D E Diagnosis
January 2012 November 2012
18. Population
•Align by diagnosis date
•Aggregate patient event networks
•Split by user-chosen intervals
A B C Diagnosis
B C E Diagnosis
A B D Diagnosis
C D E Diagnosis
19. Population
•Align by diagnosis date
•Aggregate patient event networks
•Split by user-chosen intervals
A B C Diagnosis
B C E Diagnosis
A B D Diagnosis
C D E Diagnosis
26. Evaluation
•4 Medical Scientists
•Cardiologist, ER, ER, Epidemiologist
• “Help clinicians make earlier
diagnoses”
• “Help prioritize preventative
strategies to avoid onset of HF”
27. Insights
• “Symptoms of HF are documented months to
years preceding clinical diagnosis but are also
present in non-HF patients.”
• “Rapid increase in HF symptoms are more
prominent in cases rather than in matched
controls.”
• “MatrixFlow can contribute to the further
refinement of diagnostic criteria and may allow
for earlier prediction of HF in a primary care
population.”
28. MatrixFlow
Thanks!
adam.perer@us.ibm.com
http://www.research.ibm.com/healthcare
Adam Perer and Jimeng Sun
Healthcare Analytics Research Group
IBM T.J. Watson Research Center