Mining co-occurrence and sequence patterns from cancer diagnoses in New York State

Area of Science:

• Oncology
• Data Science
• Public Health

Background:

• Cancer diagnosis histories are complex, often involving multiple conditions.
• Understanding disease co-occurrence and progression is crucial for effective patient management.

Purpose of the Study:

• To discover disease co-occurrence and sequence patterns in cancer patients.
• To identify health disparities in cancer diagnosis patterns among diverse patient groups.
• To provide insights for clinical researchers on comorbidities and disease progression.

Main Methods:

• Utilized inpatient discharge and outpatient visit records from New York State (2011-2015).
• Generated patient diagnosis sequences for seven common cancer types.
• Applied Apriori algorithm for frequent disease co-occurrence mining and cSPADE for sequence pattern discovery.

Main Results:

• Identified distinct co-occurrence and sequence patterns for different cancer types.
• Observed significant disparities in patterns across age groups and claim types (inpatient, outpatient, ED, ambulatory surgery).
• Found gender-specific disparities in cancer types, with higher pattern support generally in males.

Conclusions:

• The adopted methods successfully generated clinically meaningful disease patterns.
• These patterns highlight potential comorbidities and disease progression pathways.
• Results underscore the importance of analyzing patient data for identifying health disparities.