Patient journey through cases of depression from claims database using machine learning algorithms

Area of Science:

• Health Informatics
• Computational Biology
• Data Science

Background:

• Real-world data from health insurance and hospital claims are increasingly available in Japan.
• Machine learning methods are being explored for analyzing this data, but methodologies for visualizing patient journeys remain under development.
• Existing machine learning approaches estimate data correlation structures for disease classification and prognosis prediction.

Purpose of the Study:

• To apply association analysis to real-world data to define patient journeys for depression diagnoses.
• To address the limitations of association analysis in simultaneously interpreting multiple outcome measures.
• To utilize Topological Data Analysis (TDA) Mapper for sequential interpretation of multiple indices and visual classification of depression-associated diseases.

Main Methods:

• Applied association analysis to real-world health insurance and hospital claims data to model patient journeys for depression.
• Employed Topological Data Analysis (TDA) Mapper to interpret multiple indices and visualize disease associations sequentially.
• Focused on a cohort with depression diagnoses.

Main Results:

• Developed a visual and continuous classification of diseases commonly associated with depression using TDA Mapper.
• Demonstrated the utility of TDA Mapper in overcoming the interpretability challenges of traditional association analysis for complex health data.
• Identified patterns in patient journeys related to depression diagnoses.

Conclusions:

• The TDA Mapper approach provides a novel method for visualizing and understanding complex disease associations within patient journeys.
• This visual classification can significantly contribute to precision medicine research by identifying comorbidities and informing personalized treatment strategies.
• The findings support the use of advanced data analysis techniques for leveraging real-world health data to improve patient care and pharmaceutical research.