Building knowledge for poison control: the novel pairing of communication analysis with data mining methods

Area of Science:

• Health Informatics
• Nursing Research
• Data Mining

Background:

• Clinical data repositories often lack crucial nursing-related concepts like communication.
• Implicit nursing actions, such as communication, significantly impact patient outcomes but are rarely documented.
• Existing data mining methods for nursing knowledge models are limited by incomplete data.

Purpose of the Study:

• To develop a hybrid model for predicting poison control center outcomes.
• To integrate qualitative communication analysis with data mining for enhanced predictive accuracy.
• To identify implicit and explicit factors influencing poison control center outcomes.

Main Methods:

• A three-year study combining qualitative analysis of nurse-patient communication with data mining of archived clinical data.
• Analysis of SPI-caller communication patterns in a regional poison control center.
• Application of knowledge discovery and data mining methods to predict outcomes.

Main Results:

• A novel approach pairing qualitative communication insights with quantitative data mining predictions.
• Development of a hybrid model incorporating both documented and undocumented factors.
• Laying the groundwork for future decision support interventions in poison control.

Conclusions:

• Integrating qualitative communication data with data mining improves the prediction of health outcomes.
• Addressing undocumented nursing actions is crucial for building comprehensive knowledge models.
• This hybrid approach offers a foundation for advanced decision support systems in healthcare.