Use of Artificial Intelligence in Health Care
What is Artificial intelligence And Historical Overview
The term artificial intelligence (AI) was first used in 1956 at a computer conference at Dartmouth College. Artificial intelligence has been variously defined as: the design and operation of computer systems capable of improved performance based on:
(a) experience (ie, learning)
(b) the computerization of activities that people believe involve thinking (such as problem solving and decision making)
(c) the development of computer systems that exhibit what
people describe as intelligence, or the ability to reason and learn from
experience. All three defined areas-machine learning, decision making, and
reasoning-have produced distinct lines of research.
Artificial Intelligence And Human Cognitive Model Of Learning
Typical areas of Al research include cognitive models of human learning, machine learning models, case-based learning models, and neural network research.
The Navy Center for Applied Research in Artificial Intelligence conducts advanced research in several of these fields, especially those of machine learning, sensor-based control of autonomous activity, integration of a variety of reasoning to support complex decision making, and neural networks.
Other major
centers of Al development are located at Massachusetts Institute of Technology,
the University of Georgia, and SRI International (produces commercial Al
products).
Four Areas Of Capabilities
Four capabilities have been identified for a computer to be able to produce an artificially intelligent product. First, it must be programmed with natural language processing to enable successful communication in a human language.
Second, it must have a strategy for knowledge representation so that it can store its own knowledge base as well as the information input by the user. Third, it must have programming that provides it with one or more information-processing and problem-solving strategies. Fourth, it must have machine learning strategies programmed.
The research areas in Al that hold the
most promise for nursing applications are machine learning, expert systems, and
knowledge engineering and representation. The fourth requirement needs further
definition because how a machine learns mimics a human process that may not be
known to all readers.
Machine Learning And Reasoning
Much work has been done on machine learning and reasoning in Defense Department laboratories. Machine learning requires the machine to evaluate its own performance and to change its decision-making strategies when performance success drops below pre-determined acceptability levels.
In general, this area of research focuses on pattern recognition and pattern reconstruction. Pattern recognition is a major source of human understanding, and making changes in mental protein-solving patterns is one definition of learning. Learning has been defined as adaptation to new circumstances by extrapolating the parameters of the problem and the deficiencies of the old problem-solving pattern to newly constructed patterns.
The new patterns are tested until a more successful
pattern is found. Thus, pattern recognition and elaboration is defined as the
nature of learning.
Artificial Intelligence And Nursing
When machines are programmed to recognize ineffectiveness of existing patterns and to construct and test changes in those patterns until a new pattern proves more successful, they are considered to exhibit machine learning. This area of research has produced significant new knowledge and applications in the defense industry.
Of greater interest to nursing, it also led to new understandings about human reasoning and ways to improve human thinking, problem solving, and decision making. Woolery, Grzymala Busse , Summers, & Budihardjo (1991) examined the use of machine learning for development of expert systems in nursing.
The term Al has been used to determine both expert systems and true artificially intelligent systems. The confusion stems from differences among users in the meaning of the term intelligence. The Al literature discusses two capacities of human intelligence: reasoning and learning from experience.
All expert systems reason; that is, they apply one or more problem-solving strategies to specific information provided by a user and produce expert advice (or a decision) as a product. When humans perform this process, they are using reason. Some Al researchers add the requirement for machine learning to the definition of AL.
Computer systems that are sophisticated enough to analyze their own performance and change their processing strategies in response to "experience" are said to learn. The capacity to learn is what differentiates Al from expert decision-support systems that do not achieve the level of true intelligence.
It is typical to find the terms Al and expert
system used interchangeably. However, the term Al should be restricted to
systems that both reason and learn from experience.
Artificial Intelligence And Nursing Research
In nursing the majority of publications that list Al as a search keyword address computerized nursing expert systems, which are usually clinical decision-support tools. The terms expert systems and decision-support systems are used interchangeably. Primarily, these are systems that help support decisions about nursing assessment or care planning.
Much work also has been done on nurse staffing and scheduling systems, such as the MEDICUS or GRASP systems. These are management decision support systems that could also be considered expert systems for management.
Decision-support systems may serve as
an online reference without much reasoning ability, Poison control centers use
such systems to determine the lethality and antidotes (if any) to a variety of
substances considered to be poisonous to human beings.
Artificial Intelligence Ant Problem Solving In Nursing
Other types of expert systems accept data input from the user and provide a recommended course of action based on a preprogrammed problem-solving strategy. Still others guide the user in the selection of one or more problem-solving algorithms.
The latter system may not offer action recommendations but merely
serve to support a logical, systematic approach to the user's own problem
solving abilities.
Nursing Scholarship Criteria And Artificial Intelligence
Another line of nursing scholarship in the field of Al involves knowledge engineering in nursing (Chase, 1988). Knowledge engineering is a subfield of Al that seeks to under-stand the ways in which nursing experts conceptualize and define nursing problems and how they think about developing problem-solving strategies.
Knowledge representation studies focus on a component of knowledge engineering. This field seeks methods of representing (programming and storing) information and human thinking processes in the computer.
Knowledge is
ultimately extracted from study of the ways that highly successful experts
mentally depict external reality (knowledge representation) and from study of
experts' problem-solving techniques, strategies, and approaches.
Artificial Intelligence as a Tool For Nurses
The amount and complexity of information available for clinical situations can easily exceed the ability of an unaided nurse to use that information clinically. The human mind evolved to function under relatively simple survival conditions, not to integrate multiple, highly complex, technical sources of information nor to calculate interaction effects and probable outcomes of many variables.
Unassisted, people cannot do that kind of work with an acceptable degree of consistency. Yet that level of information processing is exactly what modern science (and the US legal system) demands of nurses. When the requirements of a task exceed human performance parameters, people must have tools that expand their capabilities.
Artificial intelligence is one type of tool that can be developed to support and expand nurses cognitive abilities so that they can function in the sophisticated health care environment.
Give your opinion if have any.