 |
Description  |
|
|
REFERENCE TO MICROFICHE APPENDIX
Reference is hereby made to a microfiche appendix submitted herewith in accordance with 37 CFR 1.96(b). The appendix contains a computer program listing in the form of two microfiche having a total of 113 frames.
FIELD OF THE INVENTION
The present invention relates to the field of expert systems, and is directed to an expert system intended for use in treating various types of trauma, and in particular, orthopedic trauma.
BACKGROUND OF THE INVENTION
Artificial Intelligence (AI) is a branch of science resulting from the marriage of the cognitive and computer sciences. Computers, originally used for the manipulation of numbers (data) are now being used for the manipulation of ideas
(knowledge). Trends and solutions can be inferred by the assimilation of observed facts just as numbers are added and subtracted to produce totals. Computer systems are being developed that exhibit thought processes previously ascribed only to humans.
The study of AI leads to insight regarding the human thought processes in addition to the development of practical systems to solve problems in the workpiece, the school and the home. The "expert system" is one method of obtaining such practical
results with AI.
An expert system solves a problem through the manipulation of knowledge. The system consists of an inference engine and a knowledge base. The knowledge base is compiled from the experience of human experts in the field and encoded in a computer
language suited for the description of ideas and principles. The inference engine controls the flow of the program, tracing solutions.
The inference engine has, in recent years, become a widely available product through a number of companies, including Gold Hill Computers Inc., of Cambridge, Massachusetts; Intellicorp, of Mountain View, California; Technology Applications, Inc.,
of Jacksonville, Florida; Teknowledge Inc., of Palo Alto, California; Neuron Data Inc., of Palo Alto, California; and Texas Instruments, of Austin, Texas. Two inference engines have been disclosed in U.S. Pat. Nos. 4,658,370 to Erman et al., and
4,648,044 to Hardy et al., both assigned to Teknowledge Inc.
Expert systems recently have found use in a variety of applications, such as in agriculture, chemistry, computer design, construction, engineering, finance, management, health care, manufacturing, and others. For example, in U.S. Pat. No.
4,591,983 Bennett et al., an expert system for use in inventory control is disclosed, and U.S. Pat. Nos. 4,517,468, 4,642,782, and 4,644,479, all to Kemper et al., each disclose a diagnostic system for monitoring an industrial system, such as a steam
turbine generator power plant.
In the health care field, hospitals and medical laborities have used computers to analyze blood and run certain tests. Data bases have been established for recommending drug therapies for certain types of cancers. An expert system made by
Cardinal Systems Inc., Minneapolis, Minnesota, includes standard textbooks data, and a graphical illustration of the sympathetic nervous system, for purposes of testing a diagnosis, and recommending therapeutic drugs. Other expert diagnostic and
treatment systems are specific to a particular healthcare concern, such as, for example, a system called "Senex", specifically designed to aid in the treatment of breast cancer, and a system called "Hepatitis Assistant", designed for better diagnosis and
treatment of hepatitis patients. Other health care systems are known to address the specific fields of epilepsy, poison control, childbirth and physical rehabilitation.
Although prior art expert systems have been designed to address a relatively wide range of health care concerns, little is known to have been done in the area of treatment of physical trauma. That is, it is beleived that none of the existing
expert systems designed for health care applications have provided the ability to perform a consultation to help determine the optimal manner in which to treat a specific type of trauma. Such a system would be useful not only for suggesting a treatment,
but also for providing a consultation session between an experienced surgeon and a learning surgeon.
OBJECTS AND SUMMARY OF THE INVENTION
It is therefore an object of the present invention to provide an expert system directed to the treatment of physical trauma.
It is a further object of the present invention to provide an expert system for the specific field of orthopedic trauma.
It is a further object of the present invention to provide an expert system which provides one or more treatment recommendations based upon specific classifications of physical trauma.
It is a further object of the present invention to provide an expert system capable of providing a treatment recommendation based upon specific classes of orthopedic trauma.
These objects are achieved by placing textbook information, such as fracture classifications, in a database, and expert information concerning orthopedic fractures in a knowledge base. In use, a fracture to to be treated is classified, and
additional trauma information is obtained, along with some patient history. Initial treatment suggestions based upon the classification of the fracture are judged for appropriateness based upon supplemental clinical information, namely the expert
information in the knowledge base. During inferencing, addititional information may be requested by the computer as needed. Treatment suggestions are presented in the order of preferred use.
The expert system in accordance with the present invention provides the user with one or more suggested treatments for a patient with physical trauma. The system includes a computing device having a memory, a plurality of databases in the
memory, an application program and an inference engine program. The databases include graphic illustrations of different types of physical trauma and a knowledge base which contains treatment information. The application program is executed in the
computing device and interactively displays a series of screens, including at least some of the graphical illustrations, to elicit reponses from the user concerning the specific type of physical trauma and specific characteristics of the patient. The
inference engine program, which is also executed in the computing device, uses the knowledge base and information related to the responses elicited from the user, for selecting one or more suggested treatments. The application program presents the
suggested treatments to the user after execution of the inference engine program.
In accordance with a more specific aspect of the present invention, the physical trauma consists of orthopedic fractures.
BRIEF DESCRIPTION OF THE DRAWINGS
These and other objects, aspects and embodiments of the present invention will now be described in more detail with reference to the following drawing figures, of which:
FIG. 1 is a diagram illustrating the structure of the expert system in accordance with the present invention;
FIGS. 2 through 15 illustrate examples of screens which may be displayed in order to elicit information concerning characteristics of an orthopedic fracture and the patient, from the user of the system;
FIG. 16 illustrates a sample screen which can be used to illustrate to the user of the system the treatments for the particular orthopedic fracture and patient;
FIG. 17 illustrates a screen which can be used to illustrate to the user of the system possible complications associated with the particular fracture under study;
FIG. 18 illustrates the hierarchy of information within the knowledge base of the expert system;
FIG. 19 is a chart illustrating the forward and backward chaining employed by the inference ebgine, in accordance with the present invention; and
FIG. 20 is an illustration of a techanique for determining the optimal treatment, based on certainty factors and the hierarchy of treatment suggestions.
DETAILED DESCRIPTION OF THE INVENTION
The expert system in accordance with the present invention will be described with initial reference to FIG. 1. The expert system 1, includes an inference engine and processor 10, inference interface 11, application program 12 and application
interface 13. The inference engine and processor 10 functions as in inference engine, under the control of an inference engine program, and also executes the application program 12, when necessary to perform application program functions, under the
overall control of the inference engine. Communication between the expert system and the user is by way of a CTR/keyboard 15 and through the inference interface 11, for communicating with the inference engine and processor 10, and by way of application
interface 13, for communicating with the application program 12. The inference engine and processor 10 can be selected from one of the available expert system packages identified above, for example, the Texas Instruments Personal Consultant.TM. Plus.
An advantage of such systems is their ability to run on the commonly available 80286 or 80386 DOS-based personal computers.
The inference engine and processor 10 receives information from two data bases, namely a knowledge base 16, and a data base of working files 18 which are generated by the application program 12, based on information elicited from the user.
The knowledge base 16 includes a collection of rules and parameter descriptions, upon which one or more preferred treatment techniques are based. The information within the knowledge base is based on information from experts within the relevent
field, in this case physical trauma, and in particular, orthopedic fractures. The working files 18, comprised of files referred to as PATIENT, DISEASE, TRAUMA and TISSUE, reflect specific information about a patient, including, for example, the
patient's specific trauma, characteristics of the surrounding tissue, patient information such as height, weight, and the like, and information as to any pre-existing conditions, such as osteoporosis.
The knowledge base 16 permanently exists within the expert system, although it should be updated periodically in accordance with currently available expert knowledge. The working files 18, on the other hand, are specific to each patient and
injury, and therefore must be created with each use of the expert system. In accordance with the present invention, a procedures and classification data base 20, input-output graphics 22 and classification graphics 24, are provided for the purpose of
gathering the requisite patient and trauma information from the system user, and assembling that information into the working files 18. The procedures and classification description data base 20 includes, for each classification of trauma, such as
orthopedic fractures, one or more initially recommended treatments and a short description of the orthopedic fracture. The initial procedures are based on textbook, and perhaps expert information, and are used by the inference engine as a starting point
from which to determine a final group of recommended procedures. Also included in the data base 20 is a hierarchy of treatment procedures, a compilation of characteristics of each class of fracture, and a compilation of characteristics associated with
pre-existing diseases and other trauma that the patient may exhibit. These files will be described in more detail below.
The input-output graphics 22 are provided as part of the expert system to elicit responses to a series of questions relating to the patient and trauma. The classification graphics 24, are based on the particular classes of trauma which the
expert system addresses. In accordance with the present example, namely that of orthopedic fractures, the classification graphics 24 are based upon specific types of orthopedic fractures assembled from textbook and similar data. Two such sources of
information are M. E. Muller, M. Allgower, R. Schnieder and H. Willenegger, MANUAL OF INTERNAL FIXATION (Berlin, Heidelberg, New York: Springer-Verlag, 2nd ed. 1979), and M. E. Muller, S. Nazarian and P. Koch, CLASSIFICATION AO DES FRACTURES, TOME I:
LES OS LONGS (Berlin, Heidelberg, New York: Springer Verlag, August 1987).
The input-output graphics 22 and classification graphics 24, form the screens illustrated in FIGS. 2-17. It will be appreciated that these screens are exemplary and that other screens, which elicit the proper information, may be equally
suitable. Also, the order in which the screens are displayed may be changed as necessary. Finally, although the subject matter of the screens relates to orthopedic fractures, the same or similar techniques may be used to gather information as to other
types of trauma as well.
The application program, the knowledge base and the procedures and classification data base, produced in accordance with the specific example of the present invention described herein, are set forth in the microfiche appendix.
The operation of the expert system in accordance with the present invention will now be described with reference to FIGS. 2-17. Initially, it will be assumed that the expert system has begun to execute the application program, in a manner which
will be described in more detail below.
FIG. 2 is an illustration of the first of the screens provided by I/O graphics 22. As shown in FIG. 2, the expert system is called ORTHOPEDIC ADVISOR. The screen of FIG. 2 provides a "menu", namely a view of the primary bones in the human body,
from which the user can select the particular bone which has been fractured. This can be done through the use of any well-known input technique, such as a touch-screen input, or through the movement of a cursor (not shown) by way of a mouse or cursor
keys. The application program, provided in the microfiche appendix, supports the use of touch-screen technology and cursor movement through a mouse input or cursor keys. The user will either touch the screen at the location corresponding to the bone
under study or move the cursor to that bone, in order to input that information into the application program. In the present example, the selected bone will be the upper long bone of the leg, namely the femur.
Also provided as part of the screen shown in FIG. 2 are locations on the screen, designated by reference numerals 28, 30 and 32, commonly called "icons", which may be selected by the user in order to perform the stated fucntions. For example,
icon 28 which bears the legend "last", can be selected by the user either by touching the icon or by positioning the cursor over the icon, in order to instruct the application program to display the previous screen. Similarly, icon 30, which bears the
legend "next", may be selected by the user to select the next screen, and icon 32, which bears the legend "stop", may be selected to terminate the program at this point. Each of these icons appears in the screens shown in FIGS. 2 through 16, and since
their function is the same, further reference will not be made thereto.
The femur having been selected, the application program displays the screen illustrated in FIG. 3, either automatically upon selection of the particular bone, or in response to selection of icon 30 for the next screen. The screen illustrated in
FIG. 3 shows the selected bone, namely the femur, in isolation, and requests the user to select the portion of the femur which has been fractured. This can be accomplished by having the user either touch or move the cursor to the affected portion of the
femur 34, or by having the user touch or move the cursor to any one of three areas 36, 38 or 40, to select the proximal femur, the femoral shaft, or the distal femur, respectively. The two-digit number illustrated in the upper left-hand corner of the
areas 36, 38 or 40 designate the numerical classification of the site of the fracture, according to a coding scheme used by the procedures and classification data base 20.
After havig selected the affected portion of the relevent bone, in this example the proximal femur, the screen illustrated in FIG. 4 is displayed, and the user is requested to select the affected region of the proximal femur, namely the (i)
trochanteric region, (ii) the femoral neck, (iii) the femoral head, classifed as locations 31a, 31b and 31c, respectively. As shown in FIG. 4, three illustrations 42, 44 and 46 of the proximal femur are illustrated, each with cross hatching on the
respective region, namely the trochanteric region in illustration 42, the femoral neck in illustration 44 and the femoral head in illustration 46. In this manner, the user continues as before, by simply selecting the illustration that corresponds to the
fracture under study. In accordance with the present example, the femoral neck is selected as the region of fracture, and in response to that selection, the next screen, illustrated in FIG. 5, is displayed.
FIG. 5 illustrates three possible types of fractures of the femoral neck, namely the femoral neck in abduction (classification 31B1), the femoral neck with a vertical fracture line (classification 31B2), and the femoral neck in addition
(classification 31B3). These classifications of specific fracture types are taken from the relevent body of knowledge on the subject of orthopedic fractures, such as the treatises by M. E. Muller et al., referred to above. As before, three
illustrations 48, 50 and 52 are provided for the respective fracture classifications, in order to assist the user in selecting the appropriate classification. The user selects the type of fracture experience by the patient, which in this example is the
femoral neck with vertical fracture line (classification 31B2). In response, the screen illustrated in FIG. 6 is displayed, illustrated three different types of femoral neck fractures with vertical fracture lines, specifically, a fracture of the basilar
neck (classification 31B2.1), the medial neck (classification 31B2.2) and a subcapital fracture (classification 31B2.3). Three illustrations 54, 56 and 58 are provided in order to assist the user in selecting the appropriate type of fracture. In the
present example, the medial neck (classification 31B2.2) is selected, and in response, the screen illustrated in FIG. 7 is displayed.
The screen illustrated in FIG. 7, which is generated in accordance with the graphics information in I/O graphics 22, requests that the user provided information as to any other trauma that the patient has experienced. For example, if the patient
has lost blood, the user would select icon 60 to indicate cardiovascular trauma. In response to this selection, the screen illustrated in FIG. 8 is displayed and the user indicates the severity of blood loss, by selecting one of the choices 62. After
making such a selection, the screen illustrated in FIG. 7 is again displayed, and a legend indicating the severity of cardiovascular trauma will appear next to icon 60. The user can then select other types of trauma, if applicable. This information is
placed into the file TISSUE, in the working files 18, and is also used by the application program to arrive at an injury severity score (ISS), which is similarly placed in the file TISSUE. The calculation of the ISS is based on the technique disclosed
in the article by the American College of Surgeons Committee on Trauma, entitled "Field Catagorization of Trauma Patients and Hospital Trauma Index", BULLETIN OF THE AMERICAN COLLEGE OF SURGEONS, Vol. 65, February 1980, pp. 28-33.
The next screen displayed is illustrated in FIG. 9 in which the user specifies the damage to the soft tissue surrounding the fracture and the open grade. This information also goes to the file TISSUE.
The screen illustrated in FIG. 10 is then displayed and the user indicates the patient's weight, age, height and sex. In this example, the patient weighs 60 kilograms, is 35 years old, is 170 centimeters tall and is male. This screen readily
lends itself to touch-screen applications, but can also be used with cursor movement as well. The next screen, shown in FIG. 11, requests the user to input information concerning the patient's occupation or lifestyle, in the context of the injury, i.e.,
whether the patient is sedentary, active or vigorous. In this example, the patient is considered "active". The information elicited from the user by the screens shown in FIGS. 10 and 11 is placed in the PATIENT file in the working file 18.
The next screen, shown in FIG. 12, requests the user to indicate any pre-existing illnesses that the patient might have, or other treatment concerns about the patient, since such considerations could affect the patient's ability to tolerate
surgery, to heal properly, to remain convalescent (recumbency), or to follow instructions, for example. Six icons 64 designate selections of the following treatment concerns: surgery, healing, recumbency, patient reliability, bone quality and specific
diseases. The first five of these treatment concerns are used by the application program to enter information, if applicable, into the file DISEASE in the working files 18. The sixth health concern, namely SPECIFIC DISEASES, elicits information from
the user to determine whether any of the patient's pre-existing diseases would cause one of the first five health concerns, and that information would be placed into the file DISEASE, as well.
Specifically, when the user selects SPECIFIC DISEASES as a treatment concern, the application program displays the screen illustrated in FIG. 13 and requests that the user indicate, using icons 66, the physiological system associated with the
patient's specific illness. In this example, the cardiovascular system is selected, and in response to that selection, a menu 68, FIG. 14, is displayed, and the user indicates one of the particular types of specific cardiovascular diseases. In this
case, vascular insufficiently is selected. In response to this selection, the application program inquires of the procedures and classification data base 20 to see what specific treatment concerns are caused by vascular insufficiency. Using the DISEASE
data base in the procedures and classification data base 20 (a copy of which is provided in the microfiche appendix), it is determined that vascular insufficiency causes a surgery concern with a 50 percent certainty factor. Thus, in response to
selecting vascular insufficiency, the application program displays the screen illustrated in FIG. 15, to inform the user of the surgery concern. Alternatively, any of the surgery, healing, recumbency, reliability and bone quality concerns can be input
directly through the use of the first five icons 54, FIG. 12.
After entering the information concerning treatment concerns, the user can now instruct the expert system to proceed with the consultation, by selecting the NEXT icon in FIG. 15. In response, the inference engine 14, FIG. 1, applies the rules of
the knowledge base 16, to the information contained within the working files 20 concerning the specifics of the patient and the orthopedic fracture. In the event that further information is required, for a particular set of inputs, the inference engine
may generate one or more further inquires through the expert system interface 11, in order to gather the additional data from the user. For example, the expert system may inquire as to whether the injury resulted from a simple fall, from which the
inference engine might infer that the patient is osteoportic, if such information is considered to be important.
When the inference engine has gathered all the necessary information, it completes its tasks by achieving certain "goals", and then returns temporary control to the application program. A screen is then displayed, as illustrated in FIG. 16, and
includes a table 70 which shows the classification of the fracture, and selected chatacteristics of the patient and trauma. In addition, four choices 72, 74, 76 and 78 are displayed as the treatments selected by the inference engine, for this specific
case. These treatments may be ordered in terms of the most to least highly suggested, either by positional order, as displayed, or by a numerical or alphabetic indication by each treatment. In this case, the suggested procedures for this patient are an
A-frame, a hip prosthesis, a DHS.TM. implant and lag screws. Additionally, the user at this point, by selecting a particular one of the recommended treatments, for example the DHS.TM. implant (the treatment indicated by the reference numeral 76), the
application program will actually show the DHS.TM. implant 80 in place.
Finally, when the user is finished with the screen shown in FIG. 16, a further screen, illustrated in FIG. 17, can be displayed to illustrate to the user a statistical summary of complications experienced with patients with similar fractures, if
such a data is available.
The program flow will now be described with reference to an example which uses an inference engine developed by Texas Instruments, called Personal Consultant.TM. Plus, Version 2.0. It will be appreciated, however, that other inference engines,
such as those associated with some of the above-mentioned commercially available expert systems could be used as well. Reference will also be made to the following programs and data bases:
______________________________________ Program or Data Base Microfiche Appendix ______________________________________ Application Program Pages 1-70 Knowledge Base Rules and Pages 71-93 Parameter Descriptions Procedure Hierarchy Page 94
Initial Procedure Suggestions Pages 95-96 (97 not used) Classification Descriptions Pages 98-102 Batch Files Pages 103-105 Preexisting Diseases Page 106 Trauma Descriptions Page 107 Classification Expansions Pages 108-109
______________________________________
The application program, at pages 1 through 70 of the microfiche appendix, generally corresponds to the application program 12 of FIG. 1. It is written in MICROSOFT C programming language for use on IBM AT.RTM. compatible computers, and makes
use of a data base program called BTRIEVE, by SoftCraft, of Austin, Texas, and of a graphics program called ESSENTIAL GRAPHICS, Version 1.5, by Essential Software, Maplewood, New Jersey.
The Knowledge Base Rules and Parameter Descriptions at pages 71 through 93 of the microfiche appendix, form the knowledge base 16 of FIG. 1. The Procedure Hierarchy, Initial Procedure Suggestions, Classification Descriptions, Preexisting
Diseases, Trauma Descriptions and Classification Expansion databases form the database 20 of FIG. 1.
To initiate the program the user will type the command "AOPC", which will call the batch file "AOPC.BAT" found in the batch file listings in the microfiche appendix. That batch file in turn causes the inference engine program to be executed, to
thereby create the inference engine, which controls all further program flow. The inference engine first requests the name of the knowledge base with which it is to work, and by responding "AOOA", the name given to the knowledge base of the present
example, the inference engine accesses the appropriate knowledge base.
The inference engine first examines a listing of parameters entitled FRAMETYPES in the knowledge base, in order to determine the structure of the knowledge base. The parameter group FRAMETYPES is set forth on page 84 of the microfiche appendix,
but the structure it implies is also shown in FIG. 18.
With reference to that figure, the rules and parameters of the knowledge base, and associated "goals", are divided into logical categories, commonly referred to as "frames ". In the present example, four such frames exist, namely, ADVISE,
PROS-CONS, PATIENT and TRAUMA. The parameter group FRAMETYPES indicates to the inference engine the hierarchy of frames, as shown in FIG. 18. The first, and highest level frame is ADVISE, followed by PROS-CONS, which in turn is followed by the PATIENT
and TRAUMA frames, which share the lowest level in the hierarchy. Also shown in FIG. 18 for the ADVISE and PROS-CONS frames are goals associated with each frame. No goals are assigned to th | | |
