Learning from critical values - adverse event identification and classification
Milena Balcerzak · 2011
Patient safety plays a crucial role in the health care industry. Information about serious adverse events comes from multiple assessments and randomized clinical trials are not optimal for detecting such rare and unexpected events. Meta-analysis of heterogeneous trial data is quite complex and the medical errors made at this stage can result in disability, decreased quality of life, or even death. Such potential negative outcomes emphasize the need for a medical decision support system, which could detect, classify and even predict adverse events. The semantic and neural-networks methods are becoming standard in most other professional industries. Why not have these artificial intelligence learning methods in place in the clinical laboratory? In one particular study, such methods were used to pilot an early detection system of unexpected patterns of occurrences of laboratory values. In this paper, the artificial intelligent algorithms based on fuzzy set theory and semantic neural networks (to fuse bio-signals and to identify adverse event) was applied.