Measurement Tools and Processes in Rehabilitation Engineering


In every engineering discipline, measurement facilitates the use of structured procedures and decision­making processes. In rehabilitation engineering, the presence of "a human," the only or major component of the system of interest, has presented a number of unique challenges with regard to measurement. This is especially true with regard to the routine processes of rehabilitation that either do or could incorporate and rely on measurements. This, in part, is due to the complexity of the human system’s architecture, the variety of ways in which it can be adversely affected by disease or injury, and the versatility in the way it can be used to accomplish various tasks of interest to an individual.

Measurement supports a wide variety of assistive device design and prescription activities undertaken within rehabilitation engineering (e. g., Leslie and Smith [1990], Webster et al. [1985], and other chapters within this section). In addition, rehabilitation engineers contribute to the specification and design of measurement instruments that are used primarily by other service providers (such as physical and occupational therapists). As measurements of human structure, performance, and behavior become more rigorous and instruments used have taken advantage of advanced technology, there is also a growing role for rehabilitation engineers to assist these other medical professionals with the proper application of measurement instruments (e. g., for determining areas that are most deficient in an individual’s perfor­mance profile, objectively documenting progress during rehabilitation, etc.). This is in keeping with the team approach to rehabilitation that has become popular in clinical settings. In short, the role of measurement in rehabilitation engineering is dynamic and growing.

In this chapter, a top-down overview of measurement tools and processes in rehabilitation engineering is presented. Many of the measurement concepts, processes, and devices of relevance are common to applications outside the rehabilitation engineering context. However, the nature of the human population with which rehabilitation engineers must deal is arguably different in that each individual must be assumed to be unique with respect to at least a subset of his or her performance capacities and/or structural parameters; i. e., population reference data cannot be assumed to be generally applicable. While there are some exceptions, population labels frequently used such as "head-injured" or "spinal cord — injured" represent only a gross classification that should not be taken to imply homogeneity with regard to parameters such as range of motion, strength, movement speed, information processing speed, and other performance capacities. This is merely a direct realization that many different ways exist in which the human system can be adversely affected by disease or injury and recognition of the continuum that exists with regard to the degree of any given effect. The result is that in rehabilitation engineering, compared with standard human factors design tasks aimed at the average healthy population, many measurement values must be acquired directly for the specific client.

Measurement in the present context encompasses actions that focus on (1) the human (e. g., structural aspects and performance capacities of subsystems at different hierarchical levels ranging from specific neuromuscular subsystems to the total person and his or her activities in daily living, including work),

Assistive devices (e. g., structural aspects and demands placed on the human), (3) tasks (e. g., distances between critical points, masses of objects involved, etc.), and (4) overall systems (e. g., performance achieved by a human-assistive device-task combination, patterns of electrical signals representing the timing of muscle activity while performing a complex maneuver, behavior of an individual before and after being fitted with a new prosthetic devices, etc). Clearly, an exhaustive treatment is beyond the scope of this chapter. Measurements are embedded in every specialized subarea of rehabilitation engineering. However, there are also special roles served by measurement in a broader and more generic sense, as well as principles that are common across the many special applications. Emphasis here is placed on these.

There is no lack of other literature regarding the types of measurement outlined to be of interest here and their use. However, it is diffusely distributed, and gaps exist with regard to how such tools can be integrated to accomplish goals beyond simply the acquisition of numeric data for a given parameter. With rapidly changing developments over the last decade, there is currently no comprehensive source that describes the majority of instruments available, their latest implementations, procedures for their use, evaluation of effectiveness, etc. While topics other than measurement are discussed, Leslie and Smith [1990] produced what is perhaps the single most directly applicable source with respect to rehabilitation engineering specifically, although it too is not comprehensive with regard to measurement, nor does it attempt to be.

Fundamental Principles

Naturally, the fundamental principles of human physiology manifest themselves in the respective sensory, neuromuscular, information-processing, and life-sustaining systems and impact approaches to measure­ment. In addition, psychological considerations are vital. Familiarization with this material is essential to measurement in rehabilitation; however, treatment here is far beyond the scope of this chapter. The numerous reference works available may be most readily found by consulting relevant chapters in this Handbook and the works that they reference. In this section, key principles that are more specific to measurement and of general applicability are presented.

Structure, Function, Performance, and Behavior

It is necessary to distinguish between structure, function, performance, and behavior and measurements thereof for both human and artificial systems. In addition, hierarchical systems concepts are necessary both to help organize the complexity of the systems involved and to help understand the various needs that exist.

Structural measures include dimensions, masses (of objects, limb segments), moments of inertia, circumferences, contours, compliances, and any other aspects of the physical system. These may be considered hierarchically as being pertinent to the total human (e. g., height, weight, etc.), specific body segments (e. g., forearm, thigh, etc.), or components of basic systems such as tendons, ligaments and muscles.

Function is the purpose of the system of interest (e. g., to move a limb segment, to communicate, to feed and care for oneself). Within the human, there are many single-purpose systems (e. g., those that function to move specific limb segments, process specific types of information, etc.). As one proceeds to higher levels, such as the total human, systems that are increasingly more multifunctional emerge. These can be recognized as higher-level configurations of more basic systems that operate to feed oneself, to conduct personal hygiene, to carry out task of a job, etc. This multilevel view of just functions begins to help place into perspective the scope over which measurement can be applied.

In rehabilitation in general, a good deal of what constitutes measurement involves the application of structured subjective observation techniques (see also the next subsection) in the form of a wide range of rating scales [e. g., Fuhrer, 1987; Granger & Greshorn, 1984; Potvin et al., 1985]. These are often termed functional assessment scales and are typically aimed at obtaining a global index of an individual’s ability to function independently in the world. The global index is typically based on a number of items within a given scale, each of which addresses selected, relatively high-level functions (e. g., personal hygiene, mobility, etc.). The focus of measurement for a given item is often in estimate of the level of independence or dependence that the subject exhibits or needs to carry out the respective function. In addition, inventories of functions that an individual is able or not able to carry out (with and without assistance) are often included. The large number of such scales that have been proposed and debated is a consequence of the many possible functions and combinations thereof that exist on which to base a given scale. Functional assessment scales are relatively quick and inexpensive to administer and have a demonstrated role in rehabilitation. However, the nature and levels of measurements obtained are not sufficient for many rehabilitation engineering purposes. This latter class of applications generally begins with a function at the level and of the type used as a constituent component of functional assessment scales, considers the level of performance at which that function is executed more quantitatively, and incorporates one or more lower levels in the hierarchy (i. e., the human subsystems involved in achieving the specific functions of daily life that are of interest and their capacities for performance).

Where functions can be described and inventoried, performance measures directly characterize how well a physical system of interest executes its intended function. performance is multidimensional (e. g., strength, range, speed, accuracy, steadiness, endurance, etc.). Of special interest are the concepts of performance capacity and performance capacity measurement. Performance capacity represents the limits of a given system’s ability to operate in its corresponding multidimensional performance space. In this chapter, a resource-based model for both human and artificial system performance and measurement of their performance capacities is adopted [e. g., Kondraske, 1990, 1995]. Thus the maximum knee flexor strength available (i. e., the resource availability) under a stated set of conditions represents one unique performance capacity of the knee flexor system. In rehabilitation, the terms impairment, disability, and handicap [World Health Organization, 1980] have been prominently applied and are relevant to the concept of performance. While these terms place an emphasis on what is missing or what a person cannot do and imply not only a measurement but also the incorporation of an assessment or judgment based on one or more observations the resource-based performance perspective focuses on "what is present" or "what is right" (i. e., performance resource availability). From this perspective, an impairment can be determined to exist if a given performance capacity is found to be less than a specified level (e. g., less than 5th percentile value of a health reference population). A disability exists when performance resource insufficiency exists in a specified task.

While performance relates more to what a system can do (i. e., a challenge or maximal stress is implied), behavior measurements are used to characterize what a system does naturally. Thus a given variable such as movement speed can relate to both performance and behavior depending on whether the system (e. g.,

Human subsystem) was maximally challenged to respond "as fast as possible" (performance) or simply observed in the course of operation (behavior). It is also possible to observe a system that it is behaving

At one or more of its performance capacities (e. g., at the maximum speed possible, etc.) (see Table 145.1).

Subjective and Objective Measurement Methods

Subjective measurements are made by humans without the aid of instruments and objective measure­ments result from the use of instruments. However, it should be noted that the mere presence of an instrument does not guarantee complete objectivity. For example, the use of a ruler requires a human judgment in reading the scale and thus contains a subjective element. A length-measurement system with an integral data-acquisition system would be more objective. However, it is likely that that even this system would involve human intervention in its use, e. g., the alignment of the device and making the decision as to exactly what is to be measured with it by selection of reference points. Measures with more objectivity (less subjectivity) are preferred to minimize questions of bias. However, measurements that are intrinsically more objective are frequently more costly and time-consuming to obtain. Well-reasoned tradeoffs must be made to take advantage of the ability of a human (typically a skilled professional) to quickly "measure" many different items subjectively (and often without recording the results but using them internally to arrive at some decision).

It is important to observe that identification of the variable of interest is not influenced by whether it is measured subjectively or objectively. This concept extends to the choice of instrument used for objective measurements. This is an especially important concept in dealing with human performance and behavior, since variables of interest can be much more abstract than simple lengths and widths (e. g., coordination, postural stability, etc.) In fact, many measurement variables in rehabilitation historically have tended to be treated as if they were inextricably coupled with the measurement method, confounding debate regarding what should be measured with what should be used to measure it in a given context.

Measurements and Assessments

The basic representation of a measurement itself in terms of the actual units of measure is often referred to as the raw form. For measures of performance, the term raw score is frequently applied. Generally, some form of assessment (i. e., judgment or interpretation) is typically required. Assessments may be applied to (or, viewed from a different perspective, may require) either a single measure of groups of them. Subjective assessments are frequently made that are based on the practitioner’s familiarity with values for a given parameter in a particular context. However, due to the large number of parameters and the amount of experience that would be required to gain a sufficient level of familiarity, a more formal and objective realization of the process that takes place in subjective assessments is often employed. This process combines the measured value with objectively determined reference values to obtain new metrics, or scores, that facilitate one or more steps in the assessment process.

Measurement Tools and Processes in Rehabilitation Engineering

Percent normal □

подпись: percent normal □

(°pG

подпись:  (°pgFor aspects of performance, percent normal scores are computed by expressing subject Y’s availability of performance resource k[RAk(Y)] as a fraction of the mean availability of that resource in a specified reference population [RAk(pop)]. Ideally, the reference population is selected to match the characteristics of the individual as closely as possible (e. g., age range, gender, handedness, etc.).

(145.1)

Aside from the benefit of placing all measurements on a common scale, a percent normal representation of a performance capacity score can be loosely interpreted as a probability. Consider grip strength as the

© 2000 by CRC Press LLC

подпись: © 2000 by crc press llc

Hierarchical Level

Global/Composite

Total human

Human with artificial systems

Structure Function

Height • Multifunction, reconfigurable system

Weight • High-level functions: tasks of daily life

Postures (working, grooming, recreation, etc.)

Subjective • Functional assessment scales

And Single-number global index

Instrumented Level of indep. estimates

Methods

Performance

No single-number direct measurement is possible

•Possible models to integrate lower-level measures

Direct measurement (subjective and instrumented) of selected performance attribute for selected functions

Behavior

Subjective self — and family reports

Instrumented ambulatory

Activity monitors (selected attributes)

See notes under “function”

Dimensions

Shape

Etc.

Instrumented methods

подпись: dimensions
shape
etc.
instrumented methods

Complex Body Systems

Cognitive

Speech

Lifting, gait

Upper extremity

Cardiovascular/respiratory

Etc.

Multifunction, reconfigurable systems

System-specific functions

Function specific subjective rating scales

Often based on impairment/ disability concepts Relative metrics

Some instrumented performance capacity measures

Known also as “functional capacity” (misnomer)

Subjective and automated (objective) videotape evaluation

Instrumented measures of physical quantities vs. time (e. g., forces, angles, motions)

Electromyography (e. g., muscle timing patterns, coordination)

Basic Systems

Visual information processors

Flexors, extensors

Visual sensors

Auditory sensors

Lungs

Etc.

подпись: basic systems
visual information processors
flexors, extensors
visual sensors
auditory sensors
lungs
etc.

Dimensions

Shape

Masses

Moments of inertia

Instrumented methods

подпись: dimensions
shape
masses
moments of inertia
instrumented methods

Single function

System-specific functions

Subjective estimates by clinician for diagnostic and routine monitoring purposes

Instrumented measures of performance capacities (e. g., strength, extremes/range of motion, speed, accuracy, endurance, etc.)

Instrumented systems

Measure and log electrophysiologic biomechanical, and other variables vs. time

Post-hoc parameterization

Mechanical properties

Instrumented methods/imaging

подпись: mechanical properties
instrumented methods/imaging
Measurement Tools and Processes in Rehabilitation Engineering

Components of basic systems

Muscle

Tendon

Nerve

Etc.

Generally single-function

Component-specific functions

Difficult to assess for individual subjects

Infer from measures at “basic system level”

Direct measurement methods with lab samples, research applications

Difficult to assess for individual subject

Direct measurement methods with lab samples, research applications

Note: Structure, function, performance, and behavior are encompassed at multiple hierarchical levels. Both subjective and objective, instrumented methods of measurement are employed.

Performance resource. Assume that there is a uniform distribution of demands of demands placed on grip strength across a representative sample of tasks of daily living, with requirements ranging from zero to the value representing mean grip strength availability in the reference population. Further assuming

That grip strength was the only performance resource that was in question for subject Y (i. e., all other

Were available in nonlimiting amounts), the percent normal score would represent the probability that

A task involving grip strength, randomly selected from those which average individuals in the reference population could execute (i. e., those for which available grip strength would be adequate), could be

Successfully executed by subject Y. While the assumptions stated here are unlikely to be perfectly true, this type of interpretation helps place measurements that are most commonly made in the laboratory into daily-life contexts.

In contrast to percent normal metrics, z-scores take into account variability within the selected reference

Population. Subject Y’s performance is expressed in terms of the difference between it and the reference

Population mean, normalized by a value corresponding to one standard deviation unit (□) of the reference population distribution:

Measurement Tools and Processes in Rehabilitation Engineering

подпись: □(145.2)

It is important to note that valid z-scores assume that the parameter in question exhibit a normal distribution in the reference population. Moreover, z-scores are useful in assessing measures of structure, performance, and behavior. With regard to performance (and assuming that measures are based on a resource construct, i. e., a larger numeric value represents better performance), a z-score of zero is produced when the subject’s performance equals that of the mean performance in the reference popu­lation. Positive z-scores reflect performance that is better than the population mean. In a normal distri­bution, 68.3% of the samples fall between z-scores of -1.0 and +1.0, wile 95.4% of these samples fall between z-scores of -2.0 and +2.0. Due to variability of a given performance capacity within a healthy population (e. g., some individuals are stronger, faster, more mobile that others), a subject with a raw performance capacity score that produces a percent normal score of 70% could easily produce a z-score of -1.0. Whereas this percent normal score might raise concern regarding the variable of interest, the z-score of -1.0 indicates that a good fraction of healthy individuals exhibit lower level of performance

Capacity.

Both percent normal and z-scores require reference population data to compute. The best reference (i. e., most sensitive) is data for that specific individual (e. g., preinjury or predisease onset). In most cases, these data do not exist. However, practices such as preemployment screenings and regular checkups are beginning to provide individualized reference data in some rehabilitation contexts.

In yet another alternative, it is frequently desirable to use values representing demands imposed by tasks [RDk(task A)] as the reference for assessment of performance capacity measures. Demands on performance resources can be envisioned to vary over the time course of a task. In practice, an estimate of the worst-case value (i. e., highest demand) would be used in assessments that incorporate task demands as reference values. In one form, such assessments can produce binary results. For example, availability can be equal to or exceed demand (resource sufficiency),or it can be less than demand (resource insuf­ficiency). These rule-based assessments are useful in identifying limiting factors, i. e., those performance resources that inhibit a specified type of task from being performed successfully or that prevent achieve­ment of a higher level of performance in a given type of task.

If RAk [bject y) Rd [task aQ then

Measurement Tools and Processes in Rehabilitation Engineering

Ra [bject yQ is insufficient

These rule-based assessments represent the basic process often applied (sometimes subliminally) by experienced clinicians in making routine decisions, as evidenced by statements such as "not enough strength," "not enough stability," etc. It is natural to extend and build on these strategies for use with objective measures. Extreme care must be employed. It is often possible, for example, for an individual to substitute another performance resource that is not insufficient for one that is. Clinicians take into account many such factors, and objective components should be combined with subjective assessments that provide the required breadth that enhances validity of objective components of a given assessment.

Using the same numeric values employed in rule-based binary assessments, a preference capacity stress metric can be computed:

RDk [task A) Ra [subject Y[

100

Performance capacity stress [)□

Measurement Tools and Processes in Rehabilitation Engineering

(145.4)

 

Binary assessments also can be made using this metric and a threshold of 100%. However, the stress value provides additional information regarding how far (or close) a given performance capacity value is from the sufficiency threshold.

Measurement Objectives and Approaches Characterizing the Human System and Its Subsystems

Figure 145.1 illustrates various points at which measurements are made over the course of a disease or injury, as well as some of the purposes for which they are made. The majority of measurements made in rehabilitation are aimed at characterizing the human system.

Measurements of human structure [e. g., Pheasant, 1986] play a critical role in the design and pre­scription of components such as seating, wheelchairs, workstations, artificial limbs, etc. Just like clothing, these items must "fit" the specific individual. Basic tools such as measuring tapes and rulers are becoming supplemented with three-dimensional digitizers and devices found in computer-aided manufacturing.

Measurement Tools and Processes in Rehabilitation Engineering

FIGURE 145.1 Measurements of structure, performance, and behavior serve many different purposes at different points over the course of a disease or injury that results in the need for rehabilitation services.

Measurements of structure (e. g., limb segment lengths, moments of inertia, etc.) are also used with computer models [Vasta & Knodraske, 1995] in the process of analyzing tasks to determine demands in terms of performance capacity variables associated with basic systems such s flexors and extensors.

After nearly 50 years, during which a plethora of mostly disease — and injury-specific functional assessment scales were developed, the functional independence measure (FIM) [Hamilton et al., 1987; Keith et al., 1987] is of particular note. It is the partial result of a task-force effort to produce a systematic methodology (Uniform Data System for Medical Rehabilitation) with the specific intent of achieving standardization throughout the clinical service-delivery system. This broad methodology uses subjective judgements exclusively, based on rigorous written guidelines, to categorize demographic, diagnostic, functional, and cost information for patients within rehabilitation settings. Its simplicity to use once learned and its relatively low cost of implementation have helped in gaining a rather widespread utilization for tracking progress of individuals from admission to discharge in rehabilitation programs and evaluating effectiveness of specific therapies within and across institutions.

In contrast, many objective measurement tools of varying degrees of technological sophistication exist [Jones, 1995; Kondraske, 1995, Smith & Leslie, 1990; Potvin et al, 1985; Smith, 1995] (also see Further Information below. A good fraction of these have been designed to accomplish the same purposes as corresponding subjective methods, but with increased resolution, sensitivity, and repeatability. The intent is not always to replace subjective methods completely but to make available alternatives with the advantages noted for situations that demand superior performance in the aspects noted. There are certain measurement needs, however, that cannot be accomplished via subjective means (e. g., measurement of a human’s visual information-processing speed, which involves the measurement of times of less than 1 second with millisecond resolution). These needs draw on the latest technology in a wide variety of ways, as demonstrated in the cited material.

With regard to instrumented measurements that pertain to a specific individual, performance capacity measures at both complex body system and basic system levels (Fig. 145.1) Constitute a major area of activity. A prime example is methodology associated with the implementation of industrial lifting stan­dards [NIOSH, 1981]. Performance-capacity measures reflect the limits of availability of one or more selected resources and require test strategies in which the subject is commanded to perform at or near a maximum level under controlled conditions. Performance tests typically last only a short time (seconds or minutes). To improve estimates of capacities, multiple trials are usually included in a given "test" from which a final measure is computed according to some established reduction criterion (e. g., average across five trials, best of three trials). This strategy also tends to improve test-retest repeatability. Performance capacities associated with basic and intermediate-level systems are important because they are "targets of therapy" [Tourtellotte, 1993], i. e., the entities that patients and service providers want to increase to enhance the chance that enough will be available to accomplish the tasks of daily life. Thus measurements of baseline levels and changes during the course of a rehabilitation program provide important docu­mentation (for medical, legal, insurance, and other purposes) as well as feedback to both the rehabilitation team and the patient.

Parameters of human behavior are also frequently acquired, often to help understand an individual’s response to a therapy or new circumstance (e. g., obtaining a new wheelchair or prosthetic device). Behavioral parameters reflect what the subject does normally and are typically recorded over longer time periods (e. g., hours or days) compared with that required for a performance capacity measurement under conditions that are more representative of the subject’s natural habitat (i. e., less laboratory-like). The general approach involves identifying the behavior (i. e., an event such as "head flexion," "keystrokes," "steps," "repositionings," etc.) and at least one parametric attribute of it. Frequency, with units of "events per unit time," and time spent in a given behavioral or activity state [e. g., Gonapthy & Kondraske, 1990] are the most commonly employed behavioral metrics. States may be detected with electromyographic means or electronic sensors that respond to force, motion, position, or orientation. Behavioral measures can be used as feedback to a subject as a means to encourage desired behaviors or discourage undesirable behaviors.

Task characterization or task analysis, like the organization of human system parameters, is facilitated with a hierarchical perspective. A highly objective, algorithmic approach could be delineated for task analysis in any given situation [Imrhan, 1995; Maxwell, 1995]. The basic objective is to obtain both descriptive and quantitative information for making decisions about the interface of a system (typically a human) to a given task. Specifically, function, procedures, and goals are of special interest. Function represents the purpose of a task (e. g., to flex the elbow, to lift an object, to communicate). In contrast, task goals relate to performance, or how well the function is to be excuted, and are quantifiable (e. g., the mass of an object to be lifted, the distance over which the lift must occur, the speed at which the lift must be performed, etc.). In situations with human and artificial systems, the term overall task goals is used to distinguish between goals of the combined human-artificial system and goals associated with the task of operating the artificial system. Procedures represent the process by which goals are achieved. Characterization of procedures can include descriptive and quantitative components (e. g., location of a person’s hands at beginning and end points of a task, path in three-dimensional space between beginning and end points). Partial or completely unspecified procedures allow for variations in style. Goals and procedures are used to obtain numeric estimates of task demands in terms of the performance resources associated with the systems anticipated to be used to execute the task. Task demands are time dependent. Worst-case demands, which may occur only at specific instants in time, are of primary interest in task analysis.

Estimates of task demand can be obtained (1) direct measurement (i. e., of goals and procedures),

The use of physics-based models to map direct measurements into parameters that relate more readily to measurable performance capacities of human subsystems, or (3) inference. Examples of direct mea­surement include key dimensions and mass of objects, three-dimensional spatial locations between "beginning" and "end points" of objects in tasks involving the movement of objects, etc. Instrumentation supporting task analysis is available (e. g., load cells, video and other systems for measuring human position and orientation in real-time during dynamic activities), but it is not often integrated into systems for task analysis per se. Direct measurements of forces based on masses of objects and gravity often must be translated (to torques about a given body joint): this requires the use of static and dynamic models and analysis [e. g., Vista & Kondraske, 1995; Winter, 1990].

An example of an inferential task-analysis approach that is relatively new is nonlinear causal resource analysis (NCRA)[Kondraske, 1999; Kondraske et al., 1997; Kondraske, 1988]. This method was motivated by human performance analysis situations where direct analysis is not possible (e. g., determination of the amount of visual information-processing speed required to drive safely on a highway). Quantitative task demands, in terms of performance variables that characterize the involved subsystems, are inferred from a population data set that includes measures of subsystem performance, resource availabilities (e. g., speed, accuracy, etc.), and overall performance on the task in question. This method is based on the simple observation that the individual with the least amount of the given resource (i. e., the lowest performance capacity) who is still able to accomplish a given goal (i. e., achieve a given level of performance in the specified high-level task) provides the key clue. That amount of availability is used to infer the amount of demand imposed by the task.

The ultimate goal to which task characterization contributes is to identify limiting factors or unsafe conditions when a specific subject undertakes the task in question; this goal must not be lost while carrying out the basic objectives of task analysis. While rigorous algorithmic approaches are useful to make evident the true detail of the process, they are generally not performed in this manner in practice at present. Rather, the skill and experience of individuals performing the analysis are used to simplify the process, resulting in a judicious mixture of subjectives estimates and objective measurements. For example, some limiting factors (e. g., grip strength) may be immediately identified without measurement of the human or the task requirements because the margin between availability and demand is so great that quick subjective "measurements" followed by an equally quick "assessment" can be used to arrive at the proper conclusion (e. g., "grip strength is a limiting factor in this task").

Assistive devices can be viewed as artificial systems that either completely or partially bridge a gap between a given human (with his or her unique profile of performance capacities, i. e., available performance resources) and a particular task or class of tasks (e. g., communication, mobility, etc.). It is thus possible to consider the aspects of the device that constitute the user-device interface and those aspects which constitute, more generally, the device-task interface. In general, measurements supporting assessment of the user-device interface can be viewed to consist of (1) those which characterize the human and (2) those which characterize tasks (i. e., "operating" the assistive device). Each of these was described earlier. Measurements that characterize the device-task interface are often carried out in the context of the complete system, i. e., the human-assistive device-task combination (see next subsection).

Characterizing Overall Systems in High-Level-Task Situations

This situation generally applies to a human-artificial system-task combination. Examples include an individual using a communication aid to communicate, an individual using a wheelchair to achieve mobility, etc. Here, concern is aimed at documenting how well the task (e. g., communication, mobility, etc.) is achieved by the composite or overall system. Specific aspects or dimensions of performance associated with the relevant function should first be identified. Examples include speech, accuracy, stability, efficiency, etc. The total system is then maximally challenged (tempered by safety considerations) to operate along one or more of these dimensions of performance (usually not more than two dimensions are maximally challenged at the same time). For example, a subject with a communication device may be challenged to generate a single selected symbol "as fast as possible" (stressing speed without concern for accuracy). Speed is measured (e. g., with units of symbols per second) over the course of short trial (so as not to be influenced by fatigue). Then the "total system" may be challenged to generate a subset of specific symbols (chosen at random from the set of those available with a given device) one at a time, "as accurately as possible" (stressing accuracy while minimizing stress on speed capacities). Accuracy is then measured after a representative number of such trials are administered (in terms of "percent correct," for example). To further delineate the speed-accuracy performance envelope, "the system" may be chal­lenge to select symbols at a fixed rate while accuracy is measured. Additional dimensions can be evaluated similarly. For example, endurance (measure in units of time) can be determined by selecting an operating point (e. g., by reference to the speed-accuracy performance envelope) and challenging the total system "to communicate" for "as long as possible" under the selected speed-accuracy condition.

In general, it is more useful if these types of characterizations consider all relevant dimensions with some level of measurements (i. e., subjective or objective) than it would be to apply a high resolution, objective measurement in a process that considers only one aspect of performance.

Decision-Making Processes

Measurements that characterize the human, task, assistive device, or combination thereof are themselves only means to an end; the end is typically a decision. As noted previously, decisions are often the result of assessment processes involving one or more measurements. Although not exhaustive, many of the different types of assessments encountered are related to the following questions: (1) Is a particular aspect of performance normal (or impaired)? (2) Is a particular aspect of performance improving, stable, or getting worse? How should therapy be modified? (3) Can a given subject utilize (and benefit from) a particular assistive device? (4) Does a subject possess the required capacity to accomplish a given higher level task (e. g., driving, a particular job after a work-related injury, etc.)?

In Fig. 145.2, several of the basic concepts associated with measurement are used to illustrate how they enter into and facilitate systematic decision-making processes. The upper section shows raw score values as well as statistics for a healthy normal reference population in tabular form (left). It is difficult to reach any decision by simple inspection of just the raw performance capacity values. Tabular data are used to obtain percent normal (middle) and z-score (right) assessments. Both provide a more directly interpret­able result regarding subject A’s impairments. By examining the "right shoulder flexion extreme of motion" item in the figure, it can be seen that a raw score value corresponding to 51.2% normal yields a very large-magnitude, negative z-score (-10.4). This z-score indicates that virtually no one in the reference population would have a score this low. In contrast, consider similar scores for the "grip strength" item (56.2% normal, z-score = -1.99). On the basis of percent normal scores, it would appear that both of these resources are similarly affected, whereas the z-score basis provides a considerably different perspective due to the fact that grip strength is much more variable in healthy populations than the extreme angle obtained by a given limb segment about a joint, relatively speaking. As noted, z-scores account for this variability.

The lower section of Fig. 145.2 considers a situation in which the issue is a specific individual (subject A) considered in a specific task. Tabular data now include raw score values (which are the same as in upper section of the figure) and quantitative demands (typically worst case) imposed on the respective perfor­mance resources by task X. The lower-middle plot illustrates the process of individually assessing suffi­ciency of each performance resource in this task context using a rule-based assessment that incorporates the idea of a threshold (i. e., availability must exceed demand for sufficiency). The lower-right plot illustrates an analogous assessment process that is executed after computation of a stress metric for each of the performance capacities. Here, any demand that corresponds to more than a 100% stress level is obviously problematic. In addition to binary conclusions regarding whether a given capacity is or is not a limiting factor, it is possible to observe that of the two limiting resources (e. g., grip strength and right shoulder flexion extreme of motion), the former is more substantial. This might suggest, for example, that the task be modified so as to decrease the grip-strength demand (i. e., gains in performance capacity required would be substantial to achieve sufficiency) and the use of focused exercise therapy to increase shoulder flexion mobility (i. e., gains in mobility required are relatively small).

145.4 Current Limitations

Quality of Measurements

Key issues are measurement validity, reliability (or repeatability), accuracy, and discriminating power. At issue in terms of current limitations is not necessarily the quality of measurements but limitations with regard to methods employed to determine the quality of measurements and their interpretability.

A complete treatment of these complex topics is beyond the present scope. However, it can be said that standards are such [Potvin et al., 1985] that most published works regarding measurement instru­ments do address quality of measurements to some extent. Validity (i. e., how well does the measurement reflect the intended quantity) and reliability are most often addressed. However, one could easily be left with the impression that these are binary conditions (i. e., measurement is or is not reliable or valid), when in fact a continuum is required to represent these constructs. Of all attributes that relate to measurement quality, reliability is most commonly expressed in quantitative terms. This is perhaps because statistical methods have been defined and promulgated for the computation of so-called reliability coefficients [Winer, 1971]. Reliability coefficients range from 0.0 to 1.0, and the implication is that 1.0 indicates a perfectly reliable or repeatable measurement process. Current methods are adequate, at best, for making inferences regarding the relative quality of two or more methods of quantifying "the same thing." Even these comparisons require great care. For example, measurement instruments that have greater intrinsic resolving power have a great opportunity to yield smaller-reliability coefficients simply because they are capable of measuring the true variability (on repeated measurement) of the parameter in question within the system under test. While there has been widespread determination or reliability coefficients, there has been little or no effort directed toward determination of what value of a reliability coefficient is "good enough" for a particular application. In fact, reliability coefficients are relatively abstract to most practitioners.

© 2000 by CRC Press LLC

j

‘ Ц

‘ — o

‘ — 5

0 SO 100

Demand as Percentage of Performance Resource (l. e., "stress" on available resources)

Z-score

(number of standard deviation units from con population mean)

Subject A: Raw Score

Healthy Normal Population

Mean

S. D.

4.5

9.7

1.46

200

356.0

78.3

87

169.9

8.0

34

48.6

9.9

3.5

5.5

0.61

1.00

0.91

0.11

R. Upper Extremity Neuromotor Channel Capacity (bits/sec)

Isometric Grip Strength (N)

R. Shoulder Flexion Extreme of Motion (deg)

R. Supinator Isometric Strength (N-m)

Visual biformation Processing Speed (stimuli/s)

Visual Acuity (1/Visual angle in minutes) ♦

PERFORMANCE

RESOURCE

VARIABLES

*

R. Upper Extremity Neuromotor Channel Capacity (bits/sec)

Isometric Grip Strength (N)

R. Shoulder Flexion Extreme of Motion (deg)

R. Supinator Isometric Strength (N-m)

Visual Information Processing Speed (stlmull/s)

Visual Acuity (1/visual angle in minutes)

Subject A: Raw Score

Task X: Demand

4.5

3.5

200

256

87

91.8

34

17

3.5

2.46

1.00

0.772

“I“

40 60 80

Pe< wen tage of Normal [(value/healthy population mean) x 100]

[H Demands Availabilities

Measurement Tools and Processes in Rehabilitation Engineering
подпись: © 2000 by crc press llc
Measurement Tools and Processes in Rehabilitation Engineering
Measurement Tools and Processes in Rehabilitation Engineering

Subject’s Resource Availabilities and Task Demaids as Percentage of Normal Population Mean Availability

 

Measurement Tools and Processes in Rehabilitation Engineering Measurement Tools and Processes in Rehabilitation Engineering

FIGURE 145.2 Examples of different types of assessments that can be performed by combining performance capacity measures and reference values of different types. The upper section shows raw score values as well as statistics for a healthy normal reference population in tabular form (left). It is difficult to reach any decision by simple inspection of just the raw performance capacity values. Tabular data are used to obtain a percent normal assessment (middle) and a z-score assessment (right). Both of these provide a more directly interpretable result regarding subject A’s impairments. The lower section shows raw score values (same as in upper section) and quantitative demands (typically worst case) imposed on the respective performance resources by task X. The lower-middle plot illustrates the process of individually assessing sufficiency of each performance resource in this task context using a threshold rule (i. e., availability must exceed demand for sufficiency). The lower-right plot illustrates a similar assessment process after computation of a stress metric for each of the performance capacities. Here, any demand that corresponds to more than a 100% stress level is obviously problematic.

Methods for determining the quality of a measurement process (including the instrument, procedures, examiner, and actual noise present in the variable of interest) that allow a practitioner to easily reach decisions regarding the use of a particular measurement instrument in a specific application and limi­tations thereof are currently lacking. At the use of different measurements increases and the number of options available for obtaining a given measurement grows, this topic will undoubtedly receive additional attention. Caution in interpreting literature, common sense, and the use of simple concepts such as "I need to measure range of motion to within 2 degrees in my application" are recommended in the meantime [Mayer et al., 1997].

Standards

Measurements, and concepts with which they are associated, can contribute to a shift from experience — based knowledge acquisition to rule-based, engineering-like methods. This requires (1) a widely accepted conceptual framework (i. e. known to assistive device manufacturers, rehabilitation engineers, and other professionals within the rehabilitation community), (2) a more complete set of measurement tools that are at least standardized with regard to the definition of the quantity measured, (3) special analysis and assessment software (that removes the resistance to the application of more rigorous methods by enhanc­ing the quality of decisions as well as the speed with which they can be reached), and (4) properly trained practitioners. Each is a necessary but not sufficient component. Thus balanced progress is required in each of these areas.

Rehabilitation Service Delivery and Rehabilitation Engineering

In a broad sense, it has been argued that all engineers can be considered rehabilitation engineers who merely work at different levels along a comprehensive spectrum of human performance, which itself can represent a common denominator among all humans. Thus an automobile is a mobility aid, a telephone is a communication aid, and so on. Just as in other engineering disciplines, measurement must be recognized not only as an important end in itself (in appropriate instances) but also as an integral component or means within the overall scope of rehabilitation and rehabilitation engineering processes. The service-delivery infrastructure must provide for such means. At present, one should anticipate and be prepared to overcome potential limitations associated with factors such as third-party reimbursement for measurement procedures, recognition of equipment and maintenance costs associated with obtaining engineering-quality measurements, and education of administrative staff and practitioners with regard to the value and proper use of measurements.

Defining Terms

Behavior: A general term that relates to what a human or artificial system does while carrying out its

Function(s) under given conditions. Often, behavior is characterized by measurement of selected parameters or identification of unique system states over time.

Function: The purpose of a system. Some systems map to a single primary function (e. g., process visual

Information). Others (e. g., the human arm) map to multiple functions, although at any given time multifunction systems are likely to be executing a single function (e. g., polishing a car). Functions can be described and inventoried, whereas level of performance of a given function can be measured. Functional assessment: The process of determining, from a relatively global perspective, an individual’s

Ability to carry out tasks in daily life. Also, the result of such a process. Functional assessments typically cover a range of selected activity areas and include (at a minimum) a relatively gross indication (e. g., can or can’t do; with or without assistance) of status in each area.

Goal: A desired endpoint (i. e., result) typically characterized by multiple parameters, at least one of

Which is specified. Examples include specified task goals (e. g., move an object of specified mass from point A to point B in 3 seconds) or estimated task performance (maximum mass, range, speed of movement obtainable given a specified elemental performance resource availability pro­file), depending on whether a reverse or forward analysis problem is undertaken. Whereas function describes the general process of task, the goal directly relates to performance and is quantitative.

Limiting resource: A performance resource at any hierarchical level (e. g., vertical lift strength, knee

Flexor speed) that is available in an amount that is less than the worst-case demand imposed by a task. Thus a given resource can be "limiting" only when considered in the context of a specific task.

Overall task goals: Goals associated with a task to be executed by a human-artificial system combination

(to be distinguished from goals associated with the task of operating the artificial system).

Performance: Unique qualities of a human or artificial system (e. g., strength, speed, accuracy, endur­

Ance) that pertain to how well that system executes its function.

Performance capacity: A quantity in finite availability that is possessed by a system or subsystem,

Drawn on during tasks, and limits some aspect (e. g., speed, force, production, etc.) of a system’s ability to execute tasks, or, the limit of that aspect itself.

Performance capacity measurement: A general class of measurements, performed at different hierar­

Chical levels, intended to quantify one or more performance capacities.

Procedure: A set of constraints placed on a system in which flexibility exists regarding how a goal (or

Set of goals) associated with a given function can be achieved. Procedure specification requires specification of initial intermediate, and/or final states or conditions dictating how the goal is to be accomplished. Such specification can be thought of in terms of removing some degrees of freedom.

Structure: Physical manifestation and attributes of a human or artificial system and the object of one

Type of measurements at multiple hierarchical levels.

Style: Allowance for variation within a procedure, resulting in the intentional incomplete specification

Of a procedure or resulting from either international or unintentional incomplete specification of procedure.

Task: That which results from (1) the combination of specified functions, goals, and procedures or

(2) the specification of function and goals and the observation of procedures utilized to achieve the goals.

References

Fuhrer MJ. 1987. Rehabilitation Outcomes: Analysis and Measurement. Baltimore, Brookes.

Ganapathy G, Kondraske GV. 1990. Microprocessor-based instrumentation for ambulatory behavior monitoring. J Clin Eng 15(6):459.

Granger CV, Greshorn GE. 1984. Functional Assessment in Rehabilitation Medicine. Baltimore, Williams & Wilkins.

Hamilton BB, Granger CV, Sherwin FS, et al. 1987. A uniform national data system for medical rehabil­itation. In MJ Fuhrer (ed), Rehabilitation Outcomes: Analysis and Measurement, pp 137-147. Baltimore, Brookes.

Imrhan S. 2000. Task analysis and decomposition: Physical components. In JD Bronzino (ed), Handbook of Biomedical Engineering, 2nd ed., Boca Raton, Fla, CRC Press.

Jones RD. 1995. Measurement of neuromotor control performance capacities. In JD Bronzino (ed), Handbook of Biomedical Engineering. Boca Raton, Fla, CRC Press.

Keith RA, Granger CV, Hamilton BB, Sherwin FS. 1987. The functional independence measure: A new tool for rehabilitation. In MG Eisenberg, RC Grzesiak (eds), Advances in Clinical Rehabilitation, vol 1, pp 6-18. New York, Springer-Verlag.

Kondraske GV. 1988. Experimental evaluation of an elemental resource model for human performance. In Proceedings of the Tenth Annual IEEE Engineering in Medicine and Biology Society Conference, New Orleans, pp 1612-13.

Kondraske GV. 1990. Quantitative measurement and assessment of performance. In RV Smith, JH Leslie (eds), Rehabilitation Engineering, pp 101-125. Boca Raton, Fla, CRC Press.

Kondraske GV. 2000. A working model for human system-task interfaces. In JD Bronzino (ed), Handbook of Biomedical Engineering, 2nd ed., Boca Raton, Fla, CRC Press.

Kondraske GV, Johnston C, Pearson, A & Tarbox L. 1997. Performance Prediction and limiting resource identification with nonlinear causal resource analysis. Proceedings, 19th Annual Engineering in Medicine and Biology Society Conference, pp 1813-1816.

Kondraske GV, Vasta PJ. 2000. Measurement of information processing performance capacities. In JD Bronzino (ed), Handbook of Biomedical Engineering, 2nd ed., Boca Raton, Fla, CRC Press.

Maxwell KJ. 2000. High-level task analysis: Mental components. In JD Bronzino (ed), Handbook of Biomedical Engineering, 2nd ed., Boca Raton, Fla, CRC Press.

Mayer T, Kondraske GV, Brady Beals, S., & Gatchel RJ. 1997. Spinal range of motion: accuracy and sources of error with inclinometric measurement. Spine, 22(17), 1976-1984.

National Institute of Occupational Safety and Health (NIOSH). 1981. Work Practices Guide for Manual Lifting (DHHS Publication No. 81122). Washington, US Government Printing Office.

Pheasant ST. 1986. Bodyspace: Anthropometry, Ergonomics and Design. Philadelphia, Taylor & Francis.

Potvin AR, Tourtellotte WW, Potvin JH, et al. 1985. The Quantitative Examination of Neurologic Func­tion. Boca Raton, Fla, CRC Press.

Smith RV, Leslie JH. 1990. Rehabilitation Engineering. Boca Raton, Fla, CRC Press.

Smith SS. 2000. Measurement of neuromuscular performance capacities. In JD Bronzino (ed), Handbook of Biomedical Engineering, 2nd ed., Boca Raton, Fla, CRC Press.

Tourtellotte WW. 1993. Personal communication.

Vasta PJ, Kondraske GV. 1994. Performance prediction of an upper extremity reciprocal task using non­linear causal resource analysis. In Proceedings of the Sixteenth Annual IEEE Engineering in Med­icine and Biology Society Conference, Baltimore.

Vasta PJ, Kondraske GV. 2000. Human performance engineering: Computer based design and analysis tools. In JD Bronzino (ed), Handbook of Biomedical Engineering, 2nd ed., Boca Raton, Fla, CRC Press.

Webster JG, Cook AM, Tompkin WJ, Vanderheiden GC. 1985. Electronic Devices for Rehabilitation. New York, Wiley.

Winer BJ. 1971. Statistical Principles in Experimental design, 2d ed. New York, McGraw-Hill.

Winter DA. 1990. Biomechanics and Motor Control of Human Movement, 2d ed. New York, Wiley.

World Health Organization. 1980. International Classification of Impairments, Disabilities, and Handi­caps. Geneva, World Health Organization.

Further Information

The section of this Handbook entitled "Human Performance Engineering" contains chapters that address

Human performance modeling and measurement in considerably more detail.

Manufacturers of instruments used to characterize different aspects of human performance often provide

Technical literature and bibliographies with conceptual backgrounds, technical specifications and application

Examples. A partial list of such sources is included below. (No endorsement of products is implied.)

Baltimore Therapeutic Equipment Co.

7455-L New Ridge Road Hanover, MD 21076-3105 Http://www. bteco. com/

Chattanooga Group 4717 Adams Road Hixson, TN 37343 Http://www. chattanoogagroup. com/

Henley Healthcare 120 Industrial Blvd.

Sugarland, TX 77478 Http://www. henleyhealth. com/

Human Performance Measurement, Inc.

P. O. Box 1996

Arlington, TX 76004-1996

Http://www. flash. net/~hpm/

Lafayette Instrument 3700 Sagamore Parkway North Lafayette, IN 47904-5729 Http://www. lafayetteinstrument. com

The National Institute on Disability and Rehabilitation Research (NIDRR), part of the Department of Education, funds a set of Rehabilitation Engineering Research Centers (RERCs) and Research and Training Centers (RTCs). Each has a particular technical focus; most include measurements and mea­surements issues. Contact NIDRR for a current listing of these centers.

Measurement devices, issues, and application examples specific to rehabilitation are included in the following journals:

IEEE Transactions on Rehabilitation Engineering IEEE Service Center 445 Hoes Lane P. O. Box 1331

Piscataway, N. J. 08855-1331 Http://www. ieee. org/index. html

Journal of Rehabilitation Research and Development Scientific and Technical Publications Section Rehabilitation Research and Development Service 103 South Gay St., 5th floor Baltimore, MD 21202-4051 Http://www. vard. org/jour/jourindx. htm

Archives of Physical Medicine and Rehabilitation Suite 1310

78 East Adams Street Chicago, IL 60603-6103

American Journal of Occupational Therapy

The American Occupational Therapy Association, Inc.

4720 Montgomery Ln.,

Bethesda, MD 20814-3425 Http://www. aota. org/

Physical Therapy

American Physical Therapy Association 1111 North Fairfax St.

Alexandria, VA 22314 Http://www. apta. org/

Journal of Occupational Rehabilitation Subscription Department Plenum Publishing Corporation 233 Spring St.

New York, NY 10013 Http://www. plenum. com/

Hobson, D., Trefler, E. “Rehabilitation Engineering Technologies: Principles of Application.” The Biomedical Engineering Handbook: Second Edition.

Ed. Joseph D. Bronzino

Boca Raton: CRC Press LLC, 2000

Добавить комментарий

Ваш e-mail не будет опубликован. Обязательные поля помечены *