Abstract | ||
---|---|---|
With the explosion of data in healthcare, there is a growing need to develop intelligent methods for automatically mining and implementing analyses from these data. In clinical applications, longitudinal patient records are often stored in disparate systems or locations in a non-integreted manner, adding work to providers and researchers to effectively utilize the information. Recent work with treatment recommendation systems has begun to help clinicians overcome these challenges, but there remains a need for concise synthesis, abstraction, and presentation of this temporal information. We present a method for unsupervised classification of common co-occuring medication administration and patient surgical events in electronic medical record data using vector-space analysis and unsupervised cluster classification. This work was done independent of domain expertice and demonstrates a method of identifying co-occuring patient events. |
Year | DOI | Venue |
---|---|---|
2016 | 10.1145/2975167.2985842 | BCB |
Keywords | Field | DocType |
EMR, Unsupervised Classification, Structured Data, Vector-Space Models, Longitudinal Data | Recommender system,Health care,Data science,Abstraction,Information retrieval,Computer science,Disparate system,Medical record,Bioinformatics,Data model | Conference |
Citations | PageRank | References |
0 | 0.34 | 3 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Alexander Titus | 1 | 0 | 0.34 |
Rebecca Faill | 2 | 0 | 0.34 |
Amar K. Das | 3 | 420 | 51.09 |