We need to build a Fraud detection model - classify claims as fraudulent or not based on historical claims. Algorithms like Decision Tree, Random Forest, Logistic Regression are used here. These are supervised learning models and require pre-classified claims for training the model. On the other hand, if you don't have the pre-classification done, then we need to build a semi-supervised or unsupervised models to understand the patterns that lead to fraud. Need to look into sample data and arrive at the right algorithm that fits the data and engineer hundreds of features that potentially help classify a claim as fraud or genuine.
Once built, the model can be deployed to classify new claims on a real-time basis. It can be made available as a REST API that can be called by front-end apps before processing a claim.
Please message me to discuss further. I'm very keen on working in this project and I believe I can do a very good job.