| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Historical Perspectives |
Affiliation of the author: Laboratory of Computer Science, Massachusetts General Hospital, Boston, MA.
Correspondence and reprints: G. Octo Barnett, MD, Laboratory of Computer Science, 50 Staniford Street, Boston, MA 02114; e-mail: <obarnett{at}partners.org>.
Received for publication: 11/03/05; accepted for publication: 11/20/05.
 |
Introduction |
|---|
I would like to welcome the Public Health Service visitors to the Computer Research Study Section Meeting and to express my personal appreciation for the opportunity to lead this colloquium. It is both a privilege and a challenge to attempt to guide a discussion with such a group of participants and I am certain that it will be an informative experience for me. In the past few years, my work with the members of the Computer Study Section has greatly contributed to my understanding of the problems we are to examine today. The critical and imaginative ideas and the depth of under-standing displayed by these individuals have set a standard which continues to inspire me.
It is my hope that this colloquium will conclude with an active discussion. I hope—and expect—to be challenged at many points and feel that we shall all benefit from such an approach. It will be interesting to hear from those among us who may have different philosophies regarding the nature of the objectives and procedures with which we are concerned and, above all, from those present who have had considerable experience in many of the topics we will discuss.
This presentation is divided into four sections.
I do not plan to review in any detail the research project we have at the Massachusetts General Hospital, but it will be obvious that many of my opinions have been strongly influenced by the experiences we have had there and by the objectives and procedures we have evolved during the past several years.
Since it is my intent to be critical and honest in my role as chairman, and since I still retain the pretensions of being a scientist with some understanding of the scientific disciplines exemplified in medical and engineering research, I must express my distaste for the superficiality and distortions that characterize some of the published works and the publicity, in both the computing industry and in the areas where computers have been applied to medical research. I am concerned with the scientific integrity of the approach that allows a broad statement that the computer can do tasks X, Y, and Z when even a superficial investigation would reveal that X can be done only on a limited demonstration basis in an artificial environment, Y is being seriously considered as an area to be programmed three years hence when system "...." is fully functioning, and task Z looks like a challenging problem in information processing and wouldn't it be nice if we could get a National Institutes of Health grant to carry out a research project. The failure to discriminate between present reality and future speculations has been one of the major causes of frustration and misunderstanding in the medical applications of computer science.
|
I. Characterization of Patient Care |
|---|
 TopIntroduction I. Characterization of Patient... II. Computer Applications in...III. Significant Issues...IV. Role of Computer...V. Role of HospitalsVI. The Role of...ConclusionAppendix 1References |
|---|
During peak periods of activity [at Massachusetts General Hospital in 1965–1966], over 500 patients are in transit in the Hospital. Innumerable telephone calls are made between patient care areas, laboratories and various physicians' offices. Each time a patient is admitted, pertinent information is sent to 66 different areas in the hospital. It is estimated that on an average day over 6,000 doctors' orders are written and the Nursing Service administers over 30,000 drugs and treatments. It is also estimated that at least 50,000 separate items of information are entered into patient records each day, or almost 20,000,000 items each year. Understandably, the amount of information processing involved in this magnitude of activity has almost overwhelmed the established system of writing requisitions and reports by hand on multiple slips of paper which are messenger carried from one part of the hospital to another. The central role of information processing in medical care can also be examined from another viewpoint—one that emphasizes the fundamental characteristic of patient care. Essentially, the diagnosis and treatment process can be defined as an iterative sequence of acts of information processing. Patient care begins when the physician initiates his investigation by gathering selected historical information from the patient and performing a physical examination. He then analyzes the data in terms of his medical knowledge and formulates a hypothesis concerning the nature of the disease process. At this time, the physician decides whether further historical information is necessary or a more extensive physical examination required. If the decision is "yes," appropriate action is taken; if the decision is "no," consideration must be given as to whether any laboratory tests would be useful in accepting or rejecting the hypothesis, or in refining it. All of this cyclic activity may proceed at a very rapid rate and require only a few seconds in a patient with a severe laceration or several hours or days in a patient with an obscure set of ill-defined symptoms. In many situations, the physician may decide the appropriate course is to focus only on a certain symptom or particular disease process, omitting or postponing a complete diagnostic examination.
[...]
The implementation of a particular plan of action may be very complex and involve a large number of individuals from the professional and hospital staff, including such diverse groups as the nursing service, the radiotherapy unit, the operating room and anesthesiology staff, the pharmacy, the blood bank, the diet kitchen, the physiotherapy unit, the social service staff, etc. A characteristic feature of the "doctor's order" is this explosion into a myriad of different orders and communications. Thus, one simple order (such as the request for a fasting blood test) requires a staggering list of routines: filling out the laboratory requisition, determining the type and amount of the sample, informing the team which draws the blood, notifying the dietary department to delay feeding and then to reinstate the feeding schedule after the test sample has been taken, checking to prevent conflict of orders and to ascertain that the patient will be available for the collection of the sample, collecting the sample, transporting the sample with the requisition to the appropriate laboratory by the message service, logging-in the sample in the laboratory, scheduling and performing the test analysis by the laboratory technician, transcribing the test results from the laboratory work sheet to the original test requisition, filing the information to ensure that the appropriate charges to the patient will be made, transporting the complete requisition with the test result to the appropriate patient care unit and, finally, sorting all the test results for a given care unit, and posting each result in the appropriate patient's medical record.
The physician must then collect the information concerning the patient's response to therapy. This information has many different forms and comes from many different sources, such as the record of vital signs, the nursing notes, and a variety of laboratory tests as well as from repeated physical examinations and interviews with the patient. Any treatment program has some of the characteristics of an experiment and involves the same iterative process of collecting information, analyzing data, formulating an hypothesis, choosing a plan of action and then recycling through the same sequence.
Such a description of the practice of medicine is clearly an oversimplification, but it emphasizes two important facts: first, that medical care is a cooperative effort, requiring the coordinated efforts of a large number of highly trained professionals, and, secondly, that the accuracy, efficiency, and rapidity of communication and information processing play a major role in determining the quality of medical care.
Much of this information processing is a component of what is termed the "medical record." Far more than a dusty volume filed in the tombs of the Records Room, the medical record is a dynamic record of the total physician-patient-hospital encounter. It includes not only the narrative account written by the physician, but also the data on test results from the laboratories, the orders written by the physician, and the record of the patient's response to a prescribed course of therapy; in effect, it is a collection of all the numerous communications and reports describing the nature and course of the disease process.
|
II. Computer Applications in Medical Communication and Information Retrieval Systems |
|---|
In recent years there has been increasing use of remote terminals which allow the user to input and retrieve data directly from the computer without any intervention on the part of a computer operator. Increasing attention is also being given to computer applications in a conversational (or interactive) mode where the computer can directly monitor input and respond immediately with appropriate data, reports or comments. This type of operation is making available new and exciting techniques in the application of computers to patient care.
|
III. Significant Issues Encountered in Research and Development |
|---|
A. Requirement for System Analysis
It is clear that no automated information-processing system can be developed or implemented until the existent system is analyzed and the objectives and procedures are well defined. Since so little attention has been paid through the years to the modus operandi involved in giving and recording medical care, the definition of the area marked for analysis is difficult indeed. Such a definition must include more than the vague generalizations commonly found in such media as advertisements and grant applications. The following are examples of generalizations which have virtually no basis in demonstrated experience.
"The advantages of the ‘.......’ will affect virtually all people and functions within the hospital through improved efficiency, cost reduction and better patient care."
"Simplicity of order entry and terminal operation provides greater accuracy and more efficient use of the nurse's time."
"The nursing staff, as a whole, will show a general tendency to a marked increase in job satisfaction and will be more eager to work in this newly created atmosphere."
Although isolated areas of hospital procedures may have been analyzed, there has been only minimal examination of the over-all structure. Such analysis is difficult and time-consuming. Most hospitals do not have clearly defined procedures and there is marked flexibility and variation even among those that have been delineated to any extent. At Massachusetts General Hospital, for example, there are three different admission offices, each employing different forms and procedures of discharging a patient. [...] It is important to remember that the tendency to be overly concerned with total system description (a tempting mirage at best) must be tempered by a realistic assessment of the essentials, at the same time avoiding entanglement with extraneous trivia.
B. Limitations of Computer Technology
It is frequently stated that application of the computer to medical care involves only simple transfer of well-defined techniques from the industrial field. This represents a gross misunderstanding of the difference between the needs of industry and the needs of hospitals. The great majority of the techniques used in the business world are those of straightforward batch-processing, involving simple processing of numerical data. [...] There is little concern with the type of interaction and communication functions necessary in the area of hospital information processing.
Because of the dissimilarity between the needs of industry to which the computer has been applied and those of the hospital with which we are concerned, computer science research in medicine is in many areas a frontier. The following are examples of pioneering research efforts in this field:
There are two other fundamental ways in which the application of computers in industry differs from that in hospitals. To begin with, the hospital user, as opposed to his counterpart in industry, is not a computer scientist or even a trained technician, but is better described as an intelligent individual whose primary concern is the use of the computer as a tool to improve patient care. Secondly, it is essential that any system useful for information processing in a hospital must not be subject to failures or downtime once it has been established and the hospital community assured of its reliability. [...]
C. Personnel Requirements
Probably the greatest single limiting factor that curtails rapid development in the application of computers to patient care is the unavailability of competent and experienced personnel. This scarcity prevails for both personnel with medical training and those with backgrounds in computer science and engineering. Research projects of this nature require the efforts of a team of individuals trained in a number of disciplines. However, it is essential that direction and leadership come from members of the hospital staff who have the experience and training to manage such an endeavor. This is due, in large part, to the existence of an attitude among hospital personnel, similar to the closed-corporation attitude of the typical trade union, of very circumscribed allegiances and patterns of communication. It is almost impossible for an individual from outside the hospital to supervise a cooperative effort involving medical and paramedical staff. In addition, as has already been pointed out, many hospitals operate on a day-to-day basis amidst customary but poorly documented procedures, and only staff members with many years of experience have an appreciation of which functions are of concern in the design of a data-processing system and which can be overlooked or investigated at a later time. Finally, a strong leadership from individuals on the hospital staff greatly facilitates the acceptance and utilization of any new innovations associated with the introduction of the computer system.
The planning for the development of the project cannot be a function of an isolated group, but must involve the participation of individuals representing the wide spectrum of disciplines found in the hospital: physicians, nurses, administrators, laboratory technicians, pharmacists, etc. I cannot emphasize strongly enough the importance of commitment on the part of the entire hospital staff, the successful interaction of all departments is a prerequisite to the development of the system.
[...]
The design and implementation of a hospital information processing system utilize the talents of several groups of individuals who would normally not be a part of the hospital environment. For example, it is very important to have systems analysts as members of the research team. Most hospital personnel have had little or no training in the critical analysis of a functioning system; few have had experience in describing all the information-flow requirements of a complex organization such as a hospital. The individual who has specialized in this type of work can make a valuable contribution to a systems project not only in supplying talent, knowledge, and experience, but also in helping the hospital staff in learning techniques of analysis.
A successful research and development project also requires the creative efforts of a large programming staff. One of the most frustrating features of any project in this field is the delay involved in developing acceptable computer programs for the available computer system. The process of preparing programs for a large time-sharing system is still more of an art than a science; this characteristic is greatly magnified in the projects dealing with hospital information systems where the application programs interact, the data base is shared by all programs, and a high degree of reliability is required.
[...]
It is questionable whether any other area of medical research has a greater need for realistic, experienced, systematic management than exists in the area of designing an automatic data-processing system in a hospital. In effect, the challenge is to work rationally in an environment where the atmosphere is neither that of a scientific enterprise nor a business endeavor. It is an exceedingly difficult task to plan amidst such uncertainty and to direct a group of such diverse backgrounds and talents. Compounding the difficulties is the fact that a system is to be implemented where not only is the implementation a research effort, but the computer system being used to implement the system is also a research project. The staff must possess a unique combination of imagination and discipline; they must have a fervor for the possibilities of the future tempered by a patient concern with the often mundane aspects of present reality. The enthusiasm generated by an exciting research effort must be guided (and sometimes restrained) by the primary concern—improved patient care.
D. The Utilization of a Hospital as a Laboratory
Large portions of the research necessary for implementing an information-processing system in the hospital can be carried out without direct involvement in active patient care. Such areas would include the development of instrumentation to facilitate the direct recording of laboratory test data or the development of a retrieval system specifically oriented toward hospital application. However, it soon becomes necessary to evolve from the isolated laboratory to the application of these procedures in the actual hospital environment. Indeed, the ultimate test of this research activity is the actual usefulness of the system in day-to-day service operation.
It took us several years to appreciate the futility of setting up demonstration projects or carrying out the research effort as a parallel procedure to the normal hospital information processing. Obviously, such methods have value in the initial stages, but are, hopefully, soon outgrown. Perhaps the principal limitation of such an approach is that parallel operation does not—and cannot—really simulate the conditions of a service operation. Secondly, if there exists a routine information system that can be depended upon, it is human nature for the staff to withhold full support and utilization of the experimental system in favor of the old-fashioned, time-tested one. Finally, the overall performance of the system and the reaction of the staff can only be adequately evaluated if the system has been used in actual service operation for a prolonged period of time.
Most projects—certainly our own—have suffered because of these limitations, for in very few cases has the computer system been adequate either in power of reliability or had appropriate and sufficient funding, to permit service operation on a significant scale for any length of time.
[...]
E. Evolutionary Approach
The innovations introduced by computer technology and the resultant reorganization and reorientation are not rapidly assimilated; hence, any significant progress in the application of computers to patient care will occur only through an evolutionary process. There has been little previous experience in this type of research and there are few laws or precedents from which to learn. It is probably true that in most research efforts, progress is realized only through a series of relative failures; nevertheless, it is frustrating to meet with repeated disappointments when the objectives are superficially so simple. Yet one must recognize the necessity of considerable operational experience before the needs under study can be adequately defined. Fortunately, in most cases, successful use of a computer system leads to a better understanding of the problem and, at the same time, stimulates a more ambitious and sophisticated approach.
Research workers in computer application areas have learned to expect that the final program will bear little resemblance to the initial idea and that each program will undergo innumerable modifications. Evolutionary development is extremely costly in terms of personnel resources and the amount of time expended. For example, research on the function we have labeled the Medication Cycle (ordering, listing and charting of medications) has deeply involved us for over three years. The programs have been rewritten twice and patched numerous times, yet the Cycle at present is little more than a very impressive demonstration and cannot handle many of the procedures or exceptions that actually occur in the normal course of events. Its limitations reflect the considerable gap between the capabilities of a demonstration program and those of an operational system powerful and reliable enough to cope with the day-to-day functional requirements. Because of this gap, we are starting again from the beginning to examine the medication problem and to respecify the flow of information and the needs to be anticipated. After three years, we think we at last understand the nature of the problem and are able to deal with it effectively.
F. Terminal Development
One of the most critical steps in the establishment of a hospital (or any other) information-processing system is the development of a method of entering and retrieving information which is satisfactory from two points of view: providing communication with the computer in a rapid and acceptable manner AND providing the user with an instrument which is relatively simple to operate.
Almost all remote-access computer systems use some form of slow-speed typewriter device. Yet this is obviously not the best choice for applications such as entering doctor's orders or retrieving lists or long laboratory reports. Other remote-access computer systems use a set of control keys that may be identified with one set of unique overlays. With a given overlay, each key represents a particular command or unit of information. This technique is useful when there is a small set of rigidly defined inputs, but is awkward when the set of possible inputs is large and not at all usable when the inputs cannot be defined in advance.
There has been little systematic development of other terminals or of alternative techniques for the entry and retrieval of information. The research that has been done has been on an isolated basis, rather than as one aspect of a coordinated program which would consider the terminal problem within the framework of the overall objectives.
G. Transferability
There is one question that is of paramount concern to any research program sponsored by the Public Health Service in the application of computers to patient care: to what degree can the results of the program be made available for use by other similar projects and, ultimately, by the hospital community at large? The matter of transferability is assuming increasing importance because of the growth of widespread interest in the problem and because of the sense of urgency on the part of a number of hospitals.
[...]
|
IV. Role of Computer Industry |
|---|
[If the results of any such research efforts] [...] are to achieve maximal utilization by other hospitals, the most likely channel for such widespread usage is through the adoption of these ideas and developments by the various computer manufacturers. This stems from the fact that the great majority of hospitals will have to depend upon computer manufacturers to supply most of the programming resources for operational service. The need for hospital research programs to carry out active communication with computer manufacturers cannot be met through the usual channels of publication, for a number of reasons: the vast amount of information to be communicated; the complexities and ramifications of this information which cannot be communicated except through hours of discussion and explanation; and the time lags and limitations imposed by the usual publication routes. In addition, a hospital investigator who has developed experience in this area has a clear responsibility for criticizing the developments that the computer manufacturers propose and undertake, since there are very few individuals in the country who can carry out such constructive criticism of a manufacturer's activities.
|
V. Role of Hospitals |
|---|
[...]
If automatic data processing is to make a significant impact on patient care, hospital administration and the medical and nursing professions must assume a major responsibility in the effort. Commitment to the project will require that part of the best talent in the hospital be assigned to the program and that staff positions be made available to data-processing specialists at salary scales much higher than is customary in a hospital. It will require the full-time participation of hospital staff who have experience in managing a research effort and who have the ability to understand and deal with the Public Health Service funding mechanisms. It will also require a willingness on the part of the hospital to gamble on a development where the economic justification is not defined.
On the basis of presently available evidence, it is very difficult to defend any large-scale hospital information-processing system with the statement that the cost of patient care will be reduced. It is almost impossible to predict the cost of a computer system since the problems are not yet clearly delineated; even where the problems are defined, there is little comparable experience in industry that would be useful in predicting costs. In addition, hospitals have very little documentation that identifies the cost of the present techniques of information processing. This cost is probably much higher than most hospital administrators appreciate and may be as much as one fourth of the total operating budget.
It would be quite erroneous to consider cost alone as the most important aspect. The decisive factor is the relation between cost of procedures and value of procedures. For example, an automatic data- processing system can provide laboratory test results to the physician faster, in a more readable format, and probably with less error. In the administration of medications, studies have shown an error rate of over 5% using manual methods of information processing. The value of reducing this error rate through the use of automated data processing is very difficult to measure in monetary terms. It is equally difficult to measure the monetary value of reducing the need for innumerable telephone calls and interruptions. A hospital cannot afford every worthy program, no matter how desirable an individual proposal may seem; in all situations, some type of priority must be established. This type of cost accounting is, at best, qualitative and it is clear that the hospitals pioneering in this area will pay a heavier cost than those which elect to develop an information-processing system several years hence. At this time there probably are not a great number of hospitals which have the imagination and resources to carry this responsibility, and it is clear that the Public Health Service will have to assume a major role in providing inspiration, leadership and financial support.
|
VI. The Role of Public Health Service |
|---|
Most of the research and demonstration projects come to this committee for scientific review. Our principal function is to determine the research merit of a particular proposal. [...] The relation between the particular research effort under study and the overall national effort must be taken into consideration: would this particular proposal significantly contribute to the broad development in this field? Indeed, is this particular area one of national importance in the area of medical care? How critical is the need for such a project—will the ultimate benefits justify the support?
[...]
It is becoming increasingly clear that these projects cannot be evaluated by the same criteria that we would use in considering clearly defined and circumscribed research projects. In particular, it is almost impossible to apply the usual standards of scientific merit, for the evaluation of a proposal in the area of hospital information processing also involves consideration of such aspects as management, hospital staff participation, availability of adequate personnel and potential value of the results.
[...] The following are, I think, justified if the National Institutes of Health makes the decision to give strong support to development in this area:
It will take at least three to five years to define a relatively large-scale hospital information system and a longer period of time for full utilization. There will be periods of discouragement, and setbacks and reformulations by the score. In most cases, particular sets of functions will be defined, tested, modified, and implemented in a stepwise fashion. The additional sophistication gained from the research efforts and the progress of computer technology itself will inevitably lead to new areas of development and new problems. In order to facilitate the evolution of the approach as well as the computer system, continuity of support is essential.
[...]
|
Conclusion |
|---|
[...] In our concern for the quality of present-day medical care, we must do more than carry out research in the etiology and treatment of disease; we must do more than build new nursing homes, clinics and hospitals; we must do more than train the highest quality of medical and paramedical personnel. These are worthy efforts, but we must also concern ourselves with the quality, efficiency, and effectiveness of the practice of medicine and the provision of medical care. This is an area which has been too long neglected by medical schools and by the National Institutes of Health.
The practice of medicine is many things. Essentially, it is an encounter on a one-to-one basis between the physician and the patient. It is an exercise in management and administration, complex enough to challenge the highest intellectual and creative abilities. But it is also, inherently, a process of information handling and a natural challenge for today's computer technology.
[...] Results from projects dealing with isolated areas of a computer information system are very encouraging. There seems to be a real potential for facilitation of routine procedures and easing the burden of paperwork currently required of staff members, particularly in the areas of laboratory test reporting and communicating bed occupancy information. From a long-range point of view, the application of computer technology to medical care offers hope for the development of new and different concepts and radically different methods of procedure that will not only contribute to the improvement of hospital administration but will facilitate medical research and greatly improve the giving of patient care.
|
Appendix 1 |
|---|
Two representative [of the six original] sections of the appendix each containing selected annotated bibliographies on the application of computers to patient care follow below).
Overview of Appendix 1
A comprehensive bibliography on the use of computer in hospital administration, patient care procedures, and medical record automation would include at lease 2,000 references depending on the limits used in defining the field. A much smaller number was selected on the basis of choosing a representative sample of the topics and projects which seem most significant. The annotations are an attempt to place the articles in perspective and represent the author's predilections. More extensive bibliographies are contained in the notes (prepared by Karl Bartscht and Richard Jelinek) of the University of Michigan Engineering Summer Conference on the "Application of Computers in Hospitals, 1966," and also in a System Development Corporation technical report by Anne Summerfield and Sally Empey entitled "Computer-Based Information System for Medicine." A bibliography on the larger topic "Computers in Medicine" has recently been prepared from Index Medicus by Arch Turner of the University of Missouri School of Medicine.
One of the difficulties in reviewing the current state of work in this field is that much of the information is contained in unpublished progress reports, proceedings of conferences that have not yet or will not be published, and internal technical reports of IBM. In addition, there is considerable work in progress, the description of which is available only through personal communication. The various individual investigators, Dr. Helen Gee, and the staff of the National Institutes of Health Computer Study Section, and IBM have been most cooperative in supplying many of the references.
E. Procedures for Retrieving Free-text Information
In any procedure which requires coding or transformation of data at the time of input, there is likely to be distortion of data or loss of information. The act of establishing a dictionary for the purpose of retrieval and processing information assumes a hypothesis about the underlying structure of that information. Most clinical research on medical record information has been done on data which are very rigidly coded, using retrieval and analysis techniques that [...] are quite limited. However, there are several projects which have developed techniques involving fewer restrictions on input and greater sophistication on retrieval.
Korein et al.45,46 have utilized the approach of fixed first, second and third-order paragraph headings which are coded for computer processing and retrieval. The content of the paragraphs is of variable length and may consist of sentences or entire paragraphs. Retrieval is accomplished through selection of specific coded [sections with free text content, or else by exact work match in a specific free text section]. Pratt and Thomas47 have described a computer-based information processing system for pathology data, using organization and classification of diagnostic data according to the "Systematized Nomenclature of Pathology" (SNOP). SNOP is a special-purpose language which, by its vocabulary, identifies all admissible primary data terms. Its organization specifies the syntactical form of any diagnostic message, first by restricting a term to a single information field and second by fixing the order in which the form information fields must be read. This system carries out a search and identification of a subset of patient records based on any logical combination of [subject matter data within a record]. Smith and Melton48 have developed a system for coding of autopsy records wherein the coding process of the diagnostic text is performed automatically by the computer. The coding uses a classification scheme to separate words into classes such as pathological processes, anatomic sites, modifiers, etiologic agents, and operations. The coding process also includes a hierarchical structure which allows retrieval at both generic and specific levels. Because there is no syntactical checking on input, errors on retrieval are possible because of grammatical complexities such as negative statements.
Lampson49 has developed a similar approach in filing and retrieving surgical pathology data in unedited, uncoded, unrestricted prose form. This technique uses a thesaurus which relates all the words having identifiable relations (either "synonym" or "inclusion") in the vocabulary of surgical pathology data. Searches are performed by simple serial matching of the data, without regard for word order, and therefore are highly dependent on the completeness of the thesaurus. In addition, false retrieval is occasionally caused by compound queries, compound diagnoses, syntax, and semantics.
At the Massachusetts General Hospital, on the Hospital Computer Project,50,51 there is a system to permit hospital personnel to define and establish private data files, to enter or change moderately large volumes of English text or coded data, and to retrieve and mathematically manipulate selected output information without the need for special assistance or any direct intervention by trained computer programmers. The design objective in constructing the information storage and retrieval programs was to facilitate Teletype console interrogation of both special research files and standard automated hospital care files (for both active and discharge patients). The data retrieval is controlled by complex logical descriptors defined at each retrieval session.
F. Application in Diagnosis and Treatment
The application of computers in medical diagnosis is the one that has received the greatest attention by both the lay public and by computer scientists. Nevertheless, there has been little operational success in the area and most of the work has been either theoretical, developmental, or sharply limited. In 1959, Ledley and Lusted52 published an examination of the problem of diagnosis from the probability approach; this discussion has significantly influenced the application of computers to medical diagnosis. Warner et al.53 have quite successfully used the probability approach (Bayes theorem) in the diagnosis of congenital heart disease; Overall and Williams54 have applied it to the diagnosis of thyroid disease; Lodwick55 has used it in the diagnosis of bone tumors. Brodman and Van Woerkom56 have applied another type of probability model to the diagnostic screening of outpatients for common disease processes and Collen57 has used a third variation in the multiphasic screening process. A number of other models of the diagnostic process have been reported but few have been tested on clinical material.
A much more successful application of computers to medical diagnosis is the automatic classification of electrocardiographic abnormalities. There have been a number of different approaches to the classification scheme; the two investigators who have most thoroughly investigated the problem are Pipberger et al. and Caceres. Pipberger et al.58 use orthogonal leads and a form of multivariant analysis. Caceres59 uses the standard 12-lead electrocardiogram and attempts to duplicate the diagnostic procedures used in clinical electrocardiography.
All applications of computers for diagnosis have been based on correlating output of the model with known clinical conditions determined by independent methods. The inherent difficulty of collecting sufficient documental and accurate clinical material has been a major limiting factor in the development of this area.
Computers have been used for treatment planning in several applications. Two of the most successful have been menu planning for patients' meals60 and the planning and execution of radiotherapy. In planning menus, the mathematical model (using techniques of linear programming) contains the dietary requirements and preferences of the patients. It has been claimed that use of the computer to plan the menus reduces the raw food costs by up to 24% with simultaneous improvement in the quality of food service and patient care.
In treatment of a tumor with radiation, the therapist must deliver an adequate dose to the tumor without destroying vital tissues that either surround the tumor or lie along the paths that the rays must travel. Because of the complexity of dose distributions when multiple sources are used, it was impossible to do more than proceed on the basis of rough estimates prior to the application of computers. At the present time, computers are used both in the computation of dosage in interstitial radiation therapy61 and in external beam therapy.62
There have been several prototype applications of computers in the management of life-threatening situations. In certain critically ill patients, it is very important to monitor a variety of physiological variables constantly, in order to establish the current status and to aid in the selection of treatments. Weil et al.63 are primarily concerned with the development of transducers and the on-line monitoring of patients critically ill with shock. Vallbona et al64 have a research effort utilizing considerable analog pre-programming that is concerned with the preparation of chronological listings and summary reports of all entries made over a prespecified period of time.
|
References |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
