Abstract
Medical information is spread into countless different data sources such as websites and databases. Some of the sources contain detailed information about a specific disease while others have general information about wellbeing and illnesses. The common factor in the sources is that they contain useful information about health related risks. By integrating the information of several data sources we find indicators and symptoms that can be used in early prediction, disease prevention and health assessment. In this paper we propose a risk mining framework for extracting and modelling health related risk patterns. We use risk mining to populate a knowledge base of health risk patterns to be used in health assessment and early prediction of health risks. Unlike in previous work done in the area of risk mining, we concentrate on risk patterns that include impact information, and not just attributes. This is crucial, as both have an important role when diagnosing patients. The proposed risk mining framework consist of four steps: (1) Risk identification, (2) Risk assessment, (3) Risk factor extraction, and (4) Risk modelling.
| Original language | English |
|---|---|
| Publication status | Published - 2010 |
| MoE publication type | Not Eligible |
| Event | Workshop on Data Mining for Healthcare Management held in conjunction with the 14th Pacific-Asia Conference on Knowledge Discovery and Data Mining, DMHM-PAKDD'10 - Hyderabad, India Duration: 21 Jun 2010 → 24 Jun 2010 |
Workshop
| Workshop | Workshop on Data Mining for Healthcare Management held in conjunction with the 14th Pacific-Asia Conference on Knowledge Discovery and Data Mining, DMHM-PAKDD'10 |
|---|---|
| Abbreviated title | DMHM-PAKDD'10 |
| Country/Territory | India |
| City | Hyderabad |
| Period | 21/06/10 → 24/06/10 |
Keywords
- risk mining
- pattern mining
- mining imbalanced data
- anomaly detection
- health knowledge bases