Step 1:
Raw Claims Data
(Patient ID, Date, Drug)

Step 2:
Mapping Drugs to Classes
(e.g., Metformin → Biguanide)

Step 3:
Transaction Sequences
(Patient-wise ordered drugs)

Step 4:
Apply SPADE Algorithm
(Vertical ID-lists + Intersections)

Step 5:
Mine Frequent Sequences
(e.g., Biguanide → Sulfonylurea → Insulin)

Step 6:
Generate Rules
(Antecedent → Consequent)

Step 7:
Predict Next Drug
(e.g., Patient on Metformin + Sulfonylurea → Predict Insulin)