Each claim is evaluated individually

ClaimScore's Scoring System

A data-driven backend analysis built with proprietary expert system AI, advanced cloud architecture and neural network machine learning.

Each claim is scored individually and tagged with justifications for score deductions.
‍
‍This system was founded on maximizing accuracy and transparency.

Claims start with a score of 1,000

Every single claim enters ClaimScore with the same clean slate.

A circular indicator completely filled in green, and the number 1000 in the center.

65+ point expert system AI review

65 points in an AI algorithm is capable of detecting 35×10¹⁸ unique fraud patterns.

Some sliders and circular indicators illustrating an expert system AI.

Score reduced for failing criteria

Criteria are weighted based on their correlation to both fraudulent and valid claims.

Tagged with deduction codes

Deduction codes provide transparency into which criteria caused the claim lose points.

A vertical bar with some tags in red, some outlined and one filled in. These represent a deduction being applied to the score.

Claims scoring ≥700 pass

If a claim's score drops below 700 they are marked as rejected.

Control studies measure accuracy

ClaimScore uses neural network machine learning to weight scoring criteria using cohorts of known valid and invalid claims. This data is segmented into learning and testing sets, in order to measure post-learning accuracy.