Medical procedure breaches detection using a fuzzy clustering approach

摘要

We present a fuzzy clustering approach for detecting flaws in the application of predefined medical protocols. In case of cardiac arrest, the outcome of the intervention is greatly influenced by the precise compliance to certain predefined procedures. We propose a clustering scheme in order to detect possible deviations from the standard protocols. Several interacting agents are employed in this regard. The fuzziness of our approach allows the discovery hybrid data which in this case may be an indication of the medical intervention quality level. We provide experiments on a dataset containing information from real patients that suffered cardiac arrest, on a synthetic dataset and on a standard dataset.