Three AI approaches for improving classification decisions concerning cytology personnel according to United States federal law (CLIA'88): An integrative study.
Item
-
Title
-
Three AI approaches for improving classification decisions concerning cytology personnel according to United States federal law (CLIA'88): An integrative study.
-
Identifier
-
AAI9908311
-
identifier
-
9908311
-
Creator
-
EL Etribi, Mohamed A.
-
Contributor
-
Adviser: George Schneller, IV
-
Date
-
1998
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Information Science | Business Administration, Management | Health Sciences, Health Care Management | Artificial Intelligence
-
Abstract
-
The use of Expert Systems (ES), Neural Networks (NN) and connectionist expert systems (ES&NN) in the delivery of health care are receiving greater attention, due to factors such as constrained budgets, staff turnover, increased malpractice suits, and increasing dependence in cytotechnologists to carry out tasks formerly performed by pathologists. This study concentrates on clinical and organizational ramifications that these intelligent systems may have on health care. This Dissertation uses specific ES, NN, and ES&NN systems in determining how such systems might be used effectively to solve a problem, and efficiently to solve fast, in the face of cost, legal and other constraints mentioned.;Clinical Laboratory testing is an important part of patient care. The Clinical Laboratory Improvement Amendment (CLIA'88) was enacted to assure that Americans receive high quality, and reliable testing in Laboratories of all types and sizes throughout the nation. CLIA'88 personnel standards require specific skills, and knowledge for Laboratory workers which vary due to the complexity of tests performed.;The study will examine the use of Expert Systems (ES) and Neural Networks (NN) technology for cytology personnel evaluations according to CLIA'88 Regulations. Specific ES, NN, and a hybrid ES&NN were developed by the author, using commercially available packages. After a sequence of preliminary tests of the three systems, the final versions were deemed valid. A questionnaire of fifteen different cases, which covered the twenty possible CLIA'88 codes (four technical supervisors, two general cytosupervisors, eight cytotechnologists, and six unqualified codes) was developed. This questionnaire was sent to three hundred evaluators in the field of medical technology. Eighty-six valid responses were received and analyzed.;The results showed that ES, NN, and ES&NN all made more accurate evaluations (the ability to correctly classify subjects into relevant codes) than Human Experts (Credentials Agency Evaluators) and Users (Clinical Laboratory Evaluators). Also, these systems are more consistent (the artificial intelligent systems are able to replicate their results) than humans. ES&NN was more accurate and consistent than ES in solving this multiple-criteria decision making problem.;The personnel evaluations of the three classes (unapproved personnel, cytotechnologists, and cytosupervisors) were obtained and analyzed for the Users and Experts. There are no significant differences for the Users and Experts in evaluating unqualified personnel and cytosupervisors. The Human Experts' evaluations of the cytotechnologists are significantly superior to Users.;The experimental results support the incorporation of ES, NN, and ES&NN systems into the present health care delivery. These preliminary studies, strongly support the efficacy of such Al systems applications in Cytology personnel evaluations. It suggests that they may have wider applications in health care.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
