Consus: Clinic Management Optimization

In order to ensure the quality of clinical documentation of patient records, a software should be developed that provides expert quality control inspectors with useful suggestions to specify and complete ICD and OPS codes for diagnoses and medical procedures. Both systematic data as well as written documentation, such as medical reports, should be considered in this process.

A probabilistic Model, developed by PSIORI, learns general coding rules from patient records in an unsupervised fashion and subsequently generates suggestions for improvements. Additionally, for a given patient record, similar records are identified and proposed for comparison by means of an artificial neural network. Furthermore, natural language processing techniques are used to find text passages that are relevant for specific codes and identify incomplete coding in medical records. All models are tested and continuously improved in cooperation with expert coders.

The final product helps experts to find relevant cases out of millions of clinical records and provides sensible suggestions for the specificity and completeness of the coding. In combination with a modern user-interface, the software accelerates the work process and thereby helps to increase the integrity and consistency of clinical patient records.