Severity Evaluation Support for Burns Unit Patients Based on Temporal Episodic Knowledge Retrieval

Severity scores are a sort of medical algorithm commonly used in medicine. In practise, physicians only use a few of them, usually internationally accepted ones involving very simple calculations. However, their daily use in critical care services gives rise to two potential problems. First, they do not always cover the particularities of the local population or a specific pathology may not be considered in the score. Second, these services (e.g. intensive care units or Burns Units) are strongly dependent on the evolution of the patients and, so the temporal component plays an essential role that should always be in mind. On the other hand, the knowledge required is at least partially present in the physician team of the medical unit due to the experience gained in treating individual patients, that is, in the form of episodic knowledge. Therefore, the use of techniques based on analogy reasoning, such as Case-Based Reasoning, would seem a suitable approach for dealing with part of this problem. In this work, we present an episodic knowledge retrieval system to support the physician in evaluating the severity patients from the temporal evolution point of view. To this end, we present different techniques for temporal retrieval based on previous works on temporal similarity. We also demonstrate the suitability of this system by applying it to a specific medical problem arising in a Burns Unit.