We apply the latest advances in domain-enhanced Artificial Intelligence to autonomously and dynamically detect FWA in claims data.
We know insurers struggle to transition from legacy coding and classification systems to modern data capture and reconciliation. That’s why Kirontech uses Medical NLP to translate medical jargon and text descriptions into standard medical codes. This fast tracks the otherwise manual process of:
Updating code sets
Mapping un-coded services and diagnosis
Classifying services into service groups
It means our coding engine can accurately process claims data with incomplete or missing diagnosis, service or drug codes. This is made possible by a combination of artificial intelligence, information engineering and advanced NLP-based techniques.
Benchmark and outlier trees both split the data into groups which hold similar practices to find outliers – if any – in each group. Benchmarks are split in a predetermined way while decision trees are grouped dynamically.
A broad statistical tool for simple analysis on a large number of practices.
Suitable for specialised, narrow analysis of selected data.
Ready to see how Kirontech could work for you?
Unsupervised learning identifies recurrent but previously unknown patterns in your data. The algorithm searches for abnormal associations without being told what to look for by a human being.
Graphs autonomously quantify the relationships and flows between entities (providers, medical conditions and patients). It learns and adapts to new relevant patterns much faster than a human being and spots complicated relationships that are not visible to the unaided eye.
Example of how our AI detects abnormal relations, such as two specialists sharing an unusual number of patients relative to peers.
Kirontech’s medical semantic network (KironMed) collects data from multiple medical sources, combining it to form one of the most comprehensive reference networks.
Our claim handling algorithms use KironMed to help them understand and enrich customer claims data.
Customers can augment Kironmed with their own reference datasets that describe specifics of their business and operational constraints.
Our built-in understanding of medical ontologies helps us navigate complex medical relationships.
Using reinforcement learning, customer feedback on recommendations reinforces desired behaviour and refines processes. Our algorithms are designed to be customizable and our team of experts work hand in hand with customers to make sure the results meet their business needs.
Representative structure of how our continous feedback mechanism works