AI-based claims analysis for health insurers
COASSESS assesses claims prior to reimbursement, flags overcharging, abnormal prices and unusual patterns, and ranks claims for review. Your investigators and managers make the decisions.
What is this landscape?The business case
of health spending adds little or nothing to health outcomes.
of health spending is lost to fraud, waste and abuse, by country and method.
Independent international estimates, not COASSESS results.
Fewer overpayments
Overcharging and failed rules are flagged before reimbursement.
Review time where it counts
Investigators start with the claims that matter most.
Stronger provider oversight
Evidence for audits and negotiations with providers.
Measured, not promised. Savings from claim audits are tracked in your COASSESS dashboard, and legitimate claims move faster.
The platform
For insurers, third-party administrators and fraud, waste and abuse (FWA) teams: from your data to a reviewed decision, in four steps.
Linked to your claims data through API connections.
Every admission and claim gets system results and rule checks.
Claims are ranked by priority and severity, then assigned.
Investigators record findings; managers accept or reject.
Automated administrative & medical review of every claim.
Does this claim break a rule — or look overcharged?
Supervised machine learning and a rulebook check every claim before reimbursement. Each rule is shown as passed or failed, so a reviewer sees exactly why a claim was flagged.
In the platform
Is anything on this invoice priced out of line?
Unsupervised machine-learning models look for abnormal costs in invoiced items. Reviewers open the invoices and price lists in one click and compare claimed costs with KEN reference prices.
In the platform
Does the cost structure of this claim look normal?
The model studies the internal structure of a claim’s costs and judges whether it is normal, after analysing a large number of historical claims.
In the platform
What should this hospitalisation cost?
At admission, the pre-claim anticipation engine estimates the expected cost — before the claim is settled. The estimate stays with the admission and the claim, so the final bill can be read against it.
In the platform
Who is connected to whom — and how often?
Insured people, agents, doctors and providers become nodes in one graph; visits, treatments, claims and shared diagnoses become links. Unusual patterns, such as unusually frequent connections, become visible for review.
In the platform
Which claims should we look at first?
Two composite scores per claim: the priority index measures how urgently it should be investigated, the severity index how serious or unusual it is.
In the platform
Trust & setup
COASSESS supports your team’s decisions. It never makes them.
Implementation follows five steps, planned with your team around your data and environment.
Data requirements
Agree what COASSESS needs from your systems.
Data validation
Exploratory analysis checks quality and yields an FWA report.
Model training
Models are trained on your data.
Deployment
The containerised platform is set up in your environment and connected to your systems.
User training
Administrators, managers and investigators get started.
Ready for use: COASSESS running in your environment, connected to your systems, with models trained on your data.
About
COVARIANCE builds AI-based solutions for healthcare and insurance organisations. COASSESS is our platform for health insurance claims.
2019
Founded by experts in health, insurance and data science
75%
of our team hold an MSc or PhD
2024: Greek Topwomen Awards
Elisavet Nika, Managing Director
2023: Best Insurtech of the Year
Startupper Awards
2022: Top Start-up in the Ecosystem
Elevate Greece National Innovation Awards
Contact
Tell us how your team reviews claims today, and we’ll show you how COASSESS fits in.
