Abstract
During the last decade, modern hospitals have witnessed a growth in the amount of information acquired, stored, and retrieved more than ever before. While aimed at helping healthcare personnel in providing care to patients, this high stream of data can also have a negative impact if not delivered in a simple and organized way. In this chapter, the authors explore the current opportunities and challenges that soft computing predictive tools face in healthcare delivery, and they then present an example of how some of these tools may contribute to the decision-making of health care providers for an important critical condition in Intensive Care Units (ICU)—septic shock. Despite current challenges, such as the availability of clean clinical data, accuracy, and interpretability, these systems will likely act to enhance the performance of a human expert and permit healthcare resources to be used more efficiently while maintaining or improving outcomes.
TopBackground
Large amounts of data are generated within the healthcare industry describing patients, hospitals resources, disease diagnosis, electronic patient records, or medical devices. These data are a very important resource to be processed and analyzed for knowledge extraction that potentially enables support for cost-savings and decision making in the organization. The key for data exploration of this type lies in data mining.
Some of the vast potential for data mining applications in healthcare can be listed as:
Key Terms in this Chapter
Evolutionary Computation (EC): AI field containing techniques that mimic processes from natural evolution and where the main concept is survival of the fittest and death of the weakest.
Artificial Neural Networks (ANN): AI method based on models of biological neurons, formed by a layered network of nodes organized on an input layer, hidden layers and an output layer.
Knowledge Discovery (KDD): Process of exploring raw data to potentially find new and ultimately understandable patterns in data.
Data Mining: Use of computational intelligence and/or statistical methods to dig out patterns in large data sets.
Swarm Intelligence (SI): AI field dedicated to the study of how social organisms perform an intelligent task through local individual interactions.
Artificial Immune Systems (AIS): AI set of methods based on the principles of natural immune systems.
Fuzzy Systems (FS): Set of AI methods based on non-crisp intervals using rules and logical connectives to establish relations between input features.
Data Mining: Use of computational intelligence and/or statistical methods to dig out patterns in large data sets.
Support Vector Machines (SVM): AI method that finds a hyperplane using support vectors and margins to separating the tuples of two classes.
Neural Networks: AI method that mimics the function of the brain, by forming a layered network of artificial neurons organized in different layers (input, hidden, and output layers).
Feature Selection: Process of finding useful features to represent the data and of removing non-relevant features containing redundant information.
Fuzzy Systems (FS): Set of AI methods based on non-crisp intervals using rules and logical connectives to establish relations between input features.
Artificial Neural Networks (ANN): AI method based on models of biological neurons, formed by a layered network of nodes organized on an input layer, hidden layers and an output layer.
Support Vector Machines (SVM): AI method that finds a hyperplane using support vectors and margins to separating the tuples of two classes.
Artificial Immune Systems (AIS): AI set of methods based on the principles of natural immune systems.
Swarm Intelligence (SI): AI field dedicated to the study of how social organisms perform an intelligent task through local individual interactions.
Evolutionary Computation (EC): AI field containing techniques that mimic processes from natural evolution and where the main concept is survival of the fittest and death of the weakest.
Neural Networks: AI method that mimics the function of the brain, by forming a layered network of artificial neurons organized in different layers (input, hidden, and output layers).
Knowledge Discovery (KDD): Process of exploring raw data to potentially find new and ultimately understandable patterns in data.
Feature Selection: Process of finding useful features to represent the data and of removing non-relevant features containing redundant information.