Related Experiment Videos
An expert-guided decision tree construction strategy: an application in knowledge discovery with medical databases
Y S Tsai1, P H King, M S Higgins
1Department of Biomedical Engineering, Vanderbilt University, Nashville, Tennessee, USA.
Summary
This study introduces an expert-guided decision tree method for knowledge discovery from electronic health records. It empowers experts to refine data analysis for better insights and identification of potential data issues.
Area of Science:
- Medical Informatics
- Data Science
- Computational Biology
Background:
- Rapid accumulation of data in electronic patient records and clinical informatics systems necessitates advanced knowledge discovery tools.
- Inter-disciplinary research demands new computational approaches for effective data management and insight generation.
Purpose of the Study:
- To propose an expert-guided decision tree construction strategy for a user-oriented knowledge discovery environment.
- To enable domain experts to influence and refine the automated decision tree building process.
Main Methods:
- Development of an expert-guided strategy for decision tree construction.
- Implementation of a system allowing experts to override inductive tree building based on expertise.
- Facilitation of expert review of decision paths to identify significant data subsets.
Main Results:
- The proposed strategy offers a more user-oriented approach to knowledge discovery from clinical data.
- Experts can effectively guide the decision tree construction, leading to more relevant findings.
- The method aids in focusing on data subsets that may represent novel discoveries or data quality issues.
Conclusions:
- Expert-guided decision trees enhance the interpretability and utility of data mining in clinical informatics.
- This approach facilitates the extraction of meaningful knowledge from large-scale electronic health record datasets.
- It provides a flexible framework for integrating expert knowledge into automated data analysis processes.