Clinical Data Mining For Physician Decision Making And Investigating Health Outcomes Methods For Prediction And Analysis PDF Download

Data mining is the process of extracting useful information from large amounts of data. Data mining has been around for a long time, and there are many multiple methods of performing data mining. However, the abundance of data that has become available in the last decade has made it possible to mine through this data to uncover important patterns and sequences. The relationship between variables and the way in which they can lead to a specific outcome is an interesting area of research. Today's healthcare industry faces a number of challenges. Providers must reduce costs, improve transparency, and improve the overall user experience. As a result of the rise of medical data, providers must leverage analytics to maximize customer data access. Additionally, patient data security is critical for regulatory compliance. Using clinical decision making with the help of data mining, analysts may now assist physicians in identifying patient concerns more effectively and in a timely manner. A physician can use data mining insights to make a more educated clinical decision and prevent patients from further clinical risks. Many data mining and machine learning techniques have been applied to several aspects of healthcare. Clinical event recognition is one of the several subfields of clinical decision making. Clinical data sequences can be used to aid in better decision making and the identification of scenarios involving patients who are at high risk of experiencing negative hospital outcomes of care. Among the negative outcomes of care include increased length of stay (LOS), negative discharge status, high mortality rate, and high cost of treatment, just to name a few instances. Our research is focused on the recognition of clinical events. We begin with some preliminary work to gain an understanding of how to use clinical data, and we then produced some statistical analyses of seasonal variations in respiratory diseases in hospital admissions, as well as demonstrated the negative impact on clinical care that occurs when a discrepancy between admission and discharge diagnosis is observed in our study. With all of the preparation work completed, our primary focus became the recognition of clinical events. In the beginning, we used an approach in which the user annotated the clinical sequence, and then we developed an Apriori-Plugin algorithm that assists in viewing the sequence of clinical events that contribute to the development of adverse clinical outcomes. Later, in order to eliminate the need for manual annotation of sequence order, we developed a Bayes-based automated extraction of clinical sequences that utilized the principles of association rule mining in conjunction with metrics such as confidence and certainty factor to extract clinical sequences. Afterward, this approach is incorporated to replace the annotation step in our prior work, which aided in the process of generating clinical sequence orders that did not require user annotation.