Understanding Healthcare Data Analytics

Sep 1, 2024

Component 24: Healthcare and Data Analytics

Unit 1: Introduction to Healthcare Data Analytics

Overview

  • Lecture A introduces the basics of healthcare data analytics.
  • Explores different types of data and technology/tools available.
  • Discusses the challenges related to big data.
  • Objectives include an overview of healthcare data analytics and the nine steps of the data analytics process.

Importance of Analytics in Healthcare

  • Peter Sondergaard (2011): "Information is the oil of the 21st century; analytics is the combustion engine."
  • Institute of Medicine (2012):
    • America’s healthcare system is complex and costly.
    • Inefficiencies and lack of focus on patient needs hinder care quality.
    • A learning healthcare system is essential for continuous improvement.

Types of Healthcare Information Systems

  • Hospitals often use various clinical systems:
    • Electronic health records (EHR)
    • Laboratory systems
    • Diagnostic imaging
    • Pharmacy, nutrition services, billing, anatomic pathology
  • These systems capture specific patient data but lack comprehensive datasets for analysis.
  • Clinical Data Warehouse: Aggregates data from multiple systems for analysis/reporting.

Data Aggregation Process

  • ETL (Extraction, Transformation, Load):
    • Extracts data from clinical systems.
    • Transforms data to synchronize formats.
    • Cleans data before loading into a clinical data warehouse.
  • Master Patient Index: Links patient identifiers across systems.

Definition of Analytics

  • Analytics vs. Statistics:
    • Analytics is broader than statistics; includes multiple BI initiatives.
  • NIST Definition (2015):
    • Discovery of meaningful patterns in data; part of the data lifecycle.

Types of Analytics

  1. Descriptive Analytics:
    • Uses BI and data mining to answer "What has happened?"
    • Presents data through charts, tables, etc.
  2. Diagnostic Analytics:
    • Advanced analytics to answer "Why did it happen?"
    • Tools: drill down techniques, data discovery, correlations.
    • Example: Kaiser Permanente's algorithm for identifying at-risk infants for sepsis.
  3. Predictive Analytics:
    • Answers "What could happen?"
    • Attributes: rapid analysis, relevance of insights, ease of use.
    • Example: Predicting sepsis risk in newborns.
  4. Prescriptive Analytics:
    • Answers "What should we do?"
    • Techniques: graph analysis, simulation, machine learning.

Steps in Data Analysis

  1. Identify the problem.
  2. Identify needed data and their locations.
  3. Develop an analysis and retrieval plan.
  4. Extract data.
  5. Check, clean, and prepare data.
  6. Analyze and interpret data.
  7. Visualize data.
  8. Disseminate knowledge.
  9. Implement knowledge into the organization.

Detailed Steps Explained

  • Step 1: Define the problem and stakeholders.
  • Step 2: Identify data elements and their sources.
  • Step 3: Create an analysis plan (consult statistician).
  • Step 4: Data extraction from systems.
  • Step 5: Ensure completeness and correctness of data.
  • Step 6: Synchronize/transform data.
  • Step 7: Import data into the analysis system.
  • Step 8: Execute the analysis plan.
  • Step 9: Communicate results clearly to stakeholders.
  • Visualizations must match data type (e.g., bar charts for categorical data).

Conclusion

  • Analytics encompasses the entire process from data collection to reporting.
  • Types of analytics: descriptive, predictive, and prescriptive.