METHOD AND SYSTEM FOR ACCURATE MEDICAL-CODE TRANSLATION
The current application is directed to methods and systems for translation of medical codes, including translation of codewords from one medical-concept code to another. The method and systems to which the current application is directed employ a multi-step translation process to translate a source codeword to a corresponding target codeword, associating the source codeword with underlying medical concepts which are, in turn, used to identify candidate target codewords of another medical-concept code. A variety of different weighting-based and filter-like criteria are then employed to select a target codeword from the candidate target codeword. The methods and systems to which the current application is directed provide for more accurate and reliable translations than would be obtained using naive, simple table-based translation.
The current application is directed to translation of medical codes and, in particular, to a method and system for accurately transmitting medical codewords from one medical-concept code to another, different, medical-concept code.
BACKGROUNDMany different medical-concept codes have been developed, including various versions of the International Statistical Classification of Diseases and Related Health Problems (“ICD”), including ICD-9 and ICD-10, as well as the systematized nomenclature of medicine (“SNOMED”). These different types of medical-concept codes provide hierarchical, alpha-numeric medical codewords for each of many different types of pathologies, diagnostics, treatments, and other medically related concepts, generally along with textural annotations and other information, much like books in libraries are encoded using the Dewey Decimal System. Medical codes are widely employed in healthcare-billing services, electronic medical records (“EMRs”), and other types of medically related information that is digitally encoded in electronic, electromagnetic, and electro-optical mass-storage devices and memories, accessed by a variety of different types of electronic data-processing systems, and displayed on various types of electronic display devices. Unfortunately, the different medical-concept codes use different alpha-numeric encodings for codewords, have different hierarchical organizations, and contain codewords that correspond to different sets of underlying concepts. It is often necessary, when processing EMRs, healthcare-billing paperwork, and other medically related information, to translate codewords from one medical-concept code to another. For example, a healthcare clinic may internally use codewords from a first medical-concept code and may need to translate these codewords to corresponding codewords of a second medical-concept code used by an insurance provider in order to facilitate processing of invoices submitted by the healthcare clinic to the insurance provider. In another example, organizations may migrate from one medical-concept code to another, the migration process involving translation of codewords stored in current EMRs and invoices to corresponding codewords of a different medical-concept code to avoid using two different types of electronic medical-data processing systems.
Unfortunately, medical-concept codes are enormous, containing many thousands of different codewords, each potentially related to numerous different underlying medical concepts. Manual translation of medical codes would be far too time-consuming and error-prone to be practical for even low-volume translation of codewords from a first medical-concept code to a second, related medical-concept code. In many cases, erroneous translation can lead to delays, unnecessary costs, and other serious and even life-threatening consequences. Because the codewords of one medical-concept code often do not conceptually align with the codewords of another medical-concept code, medical-code translation is, by nature, inexact and far from straightforward. For these reasons, medical providers, insurance companies, EMR processing companies, and many other organizations involved in medically related fields seek accurate and efficient medical-code translation to facilitate various different types of medically related tasks and operations.
SUMMARYThe current application is directed to methods and systems for translation of medical codes, including translation of codewords from one medical-concept code to another. The method and systems to which the current application is directed employ a multi-step translation process to translate a source codeword to a corresponding target codeword, associating the source codeword with underlying medical concepts which are, in turn, used to identify candidate target codewords of another medical-concept code. A variety of different weighting-based and filter-like criteria are then employed to select a target codeword from the candidate target codeword. The methods and systems to which the current application is directed provide for more accurate and reliable translations than would be obtained using naive, simple table-based translation.
The current application is directed to methods and systems for automated translation of medical codes. These methods and systems employ multi-step automated translation in which a source codeword is associated with underlying medical concepts. Candidate target codewords are then identified using the associations of the source codeword with underlying medical concepts as well as associations of the candidate target codewords with the same underlying medical concepts. A target codewords is selected from among the candidate target codewords using weighted-association comparisons, filters, and other methods.
It should be emphasized, at the onset, that the currently described methods and systems carry out real-world, important, useful tasks that result in physical transformations of electronic, electromagnetic, and electro-optical data-storage devices, electronic display of encoded information, and physical computational activities that provide tangible, real-world results. While the systems to which the current application is directed are complex computational and data-processing systems controlled by many different levels of computer instructions, these are real, tangible, physical systems that carry out real-world tasks.
An approach used in systems and methods for medical-code translation to which the current application is directed is next provided, using graphical illustrations, an example implementation using the structured query language (“SQL”) and relational databases, pseudocode, and control-flow diagrams. It should be emphasized, initially, that this discussion is not intended to cover all possible implementations or provide minute details of a particular approach within the overall approach provided below as an example. Instead, the discussion is intended to expose principles and concepts underlying many different possible implementations of the systems and methods for medical-code translation to which the current application is directed and using which particular implementations can be designed and produced.
A first step that facilitates the currently described medical-code translation process is to generate a medical-concept database. In general, generation of the medical-concept database may be carried out by using either manual, human-analyst-based methods or by using automated methods that employ natural-language processing, detailed translation rules, and inference engines. Perhaps the most productive approach is to combine both automated, semi-automated, and manual approaches to ensure that a robust, well-designed, and complete medical-concept database is prepared.
In a next step, once the medical-concept database has been prepared, an exhaustive set of associations between codewords of medical-concept codes and medical concepts stored in the medical-concept database is prepared.
As one simple example, table 1, provided below, illustrates the medical concepts associated with each of the two example codeword entries shown in
In a third step, which, in some implementations, may be combined with the second step, weights are assigned to each of the associations between medical-concept-code entries, or codewords, and medical concepts stored within a medical-concept database.
One, but by far not the only, approach to implementing method and system examples of the currently described methods and systems involves use of a relational database management system (“RDBMS”). Such systems can be managed and queried using the well-known SQL language, used below to illustrate certain portions of the example approach. In the example discussed below, rather than alpha-numeric values, the codewords are assumed to be integer values, for simplicity.
The table “code listing” 608 stores the codewords, or entries, for the medical-concept codes listed in the table “codes” 602. Each entry in table 608 includes a unique identifier of a medical-concept code 610, the codeword within the medical-concept code 612, a textural annotation for the codeword 614, such as the text annotating the two codewords shown in
The table “concepts” 620 contains the medical-concept database. Each medical concept is encoded within a row of the table. Each medical concept is encoded with a unique concept identifier 622, a textural representation of the concept 624, and potentially many additional types of information stored in column 625-626.
Finally, the table “associations” 630 stores the associations between codewords and medical-concept-database medical concepts, as illustrated in
The table “code-concept exclusions” lists pairs of codewords and concepts, each codeword represented by a pair of values in columns 710 and 712 and each concept represented by a value in column 714. Concept identifiers are obtained from the concept identifiers that uniquely identify medical concepts in column 622 of table 620 in
The table “code-code exclusions” 706 provides listings of pairs of codewords that should not represent codeword translations. The first two columns 716 and 717 specify a first codeword of a first medical-concept code and columns 718-719 specify a second codeword of a second medical-concept code. This table essentially provides a specific first-codeword-to-second-codeword exclusion filter.
The third new table illustrated in
Using the above-described tables, and the information included in them, a codeword-translation process that represents an example of the methods to which the current application is directed is next described.
Next, a number of numeric values are calculated from data stored in the intermediate tableTMP1 as well as certain of the other tables illustrated in FIGS. 6 and 7. The value “totalNum” is the total number of distinct concepts in table TMP1, or the total number of distinct medical concepts associated with either or both of the source codeword and the target codeword, and can be computed using the following SQL statement:
totalNum=SELECT COUNT (conceptNo) from TMP1
The value “numLost” is the number of concepts associated with the source codeword that are not also associated with the target codeword, and is calculated by the following SQL statement:
The value “numAssumed” is the number of medical concepts associated with the target codeword 804 but not associated with the source codeword 802, and is calculated by the following SQL statement:
The value “numMatched” is the number of distinct concepts associated both with the source codeword and the target codeword, as computed by the following SQL statement:
The total weight of associations between the source codeword and medical concepts also associated with a target codeword is computed as the value “weightMatchedF” in the following SQL statement:
The total weight of the associations from the target codeword to medical concepts also associated with the source codeword is computed as the value “weightMatchedR” by the following SQL statement:
The total weight of all associations between the source codeword and associated medical concepts and the target codeword and associated medical concepts stored in the value “weightTotal,” computed by the following SQL statement:
weightTotal=SELECT SUM (weight) FROM TMP1
The number of code/concept exclusions which are concepts associated with the target codeword that are listed as code-concept exclusions in the table “code-concepts exclusions” 714 in
The number of code/code exclusions, which are potential exclusions that would prevent translation of source codeword 802 to target codeword 804, are computed by the following SQL statement:
Finally, the number of concept pairs selected from the source codeword and target codeword that are listed as being antonyms in the table “Antonyms” 708 in
It should be noted that, in various other examples of the methods to which the current application is directed, fewer computed values can be computed and used in the matching operation. In yet alternative examples, a greater number of computed values are computed and used in the matching process. In yet additional examples, different computed values may be employed instead of in addition to, or in place of certain of the computed values discussed above.
In general, the matching operation may be considered to be a function of the above-computed values, returning a match value which indicates whether or not the source codeword and target codeword match or, in other cases, a numeric value that indicates the degree to which the target codeword matches the source codeword. In the former case, as one example, the match operation can be represented as the following function:
A slightly different match operation is provided below, in pseudocode:
The function “match” returns a Boolean value in the variable parameter “exact” that indicates whether or not the match is exact as well as a numeric value in the variable parameter “mValue” that indicates a degree of matching. The remaining parameters are the calculated values discussed above. First, on lines 6-7, if any code/code exclusions, code/concept exclusions, or antonymous concept exclusions have been discovered, then the routine “match” returns false. Otherwise, on line 10, the variable parameter “exact” is set to indicate whether the value “numMatched” is equal to the value “totalNum,” indicating that all concepts associated either with the source codeword or target codeword are associated with both the source codeword and target codeword. Next, on lines 11-12, the local variable v2 is set to the sum of numLost and numAssumed divided by totalNum. If the ratio of lost and assumed concepts to the total number of concepts is greater than a threshold value “THRESHOLD1,” the routine “match” returns false. Note that the routine “match” may return the Boolean value false even for a source codeword and target codeword that exactly match, but this rare case can be detected by inspecting the value returned in the variable parameter “exact.” In general, the routine “match” returns false when too many concepts associated either with the source codeword or target codeword are not commonly associated with the source codeword and target codeword. Next, on lines 14-15, the routine “match” computes the relative weight of common associations of the source codeword and target codeword with respect to the total weight of associations of the source codeword and target codeword. When this computed value falls below a threshold value THRESHOLD2, the routine “match” returns false. Thus, the routine “match” returns false when the relative weight of common associations with respect to the total weight of associations falls below some threshold value considered to be a minimal weight of common associations needed to match the source codeword to the target codeword. A third threshold, THRESHOLD3, a threshold for the total weight of common associations, is applied, on line 18, so that when the total weight of the common associations falls below THRESHOLD3, the routine “match” returns false. Otherwise, the summed weights of common associations is returned in the variable parameter “mValue” and the routine “match” returns the Boolean value true, on lines 19-20. Again, many alternative implementations of the matching operation are possible,
Continuing to
In the for-loop of steps 922-926, the above-described routine “match” is called for the currently considered source codeword and each of the possible translations, with any candidate translation for which the routine “match” returns a match value greater than the largest match value so far determined temporarily stored, in step 925, as the so-far-detected best candidate for translation. Upon completion of the inner for-loop of steps 922-926, the best translation identified in the inner for-loop is stored in the table in association with the currently considered source codeword. When there are more source codewords to consider, as determined in step 928, control flows back to step 921. Otherwise, the routine “generateTranslations” returns. Note that it is assumed, in this implementation, that at least one acceptable translation will be found. When this assumption is incorrect, an additional local variable may be set, in step 921, with an indication that no translation was found. That indication will be entered in the table in association with the currently considered source codeword when no translation is found in the target medical-concept code.
Not only can source codewords of one medical-concept code be translated to target codewords of another medical-concept code, the information discussed above can be used for many other medical-code-related tasks. For example, it is straightforward to generate a list of all underlying medical concepts associated with a particular codeword by, as one example, using the concisely coded SQL statement:
This is but one of many different possible examples of medical-concept-code related tasks that can be carried out using the stored information and techniques that represent examples of the methods and systems to which the current application is directed.
Although the present invention has been described in terms of particular embodiments, it is not intended that the invention be limited to these embodiments. Modifications within the spirit of the invention will be apparent to those skilled in the art. For example, medical-code-translation systems and methods can be implemented by varying any of many different design and implementation parameters, including selection of hardware platforms, operating systems, programming languages, control structures, data structures, modular organizations, and other such implementation parameters. As discussed above, while relational databases are used to provide example implementations, any of many different types of data-storage systems may be used instead of relational data systems. While, in the above example, numerous different numeric values are calculated for the source and target codewords in the match operation, in alternative examples, other numeric values may be computed and used to compute a degree of similarity or another metric returned by the match operation.
It is appreciated that the previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present disclosure. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the disclosure. Thus, the present disclosure is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A medical-code translation system comprising:
- a computer system that includes a processor, one or more electronic memories, and one or more mass-storage devices;
- a medical-concept database stored within, or accessible to, the computer system;
- and computer instructions that control operation of the medical-code-translation system, encoded in one or more of the one or more electronic memories and one or more mass-storage devices to receive and store, in memory, a source codeword from a source medical code, identify candidate target codewords of a target medical-concept code, using indications of medical concepts stored in the medical-concept database, that are associated with at least one medical concept with which the source code is associated, remove, from the candidate target codewords, excluded codewords to generate a set of remaining candidate target codewords; and select a target codeword corresponding to the source codeword that, when compared to the source codeword in a matching operation, generates a comparison metric that indicates a degree of matching with the source code greater than the degree of matching indicated by the comparison metrics generated for each of the other remaining candidate target codewords in matching operations.
2. The medical-code translation system of claim 1 wherein the medical-code translation system maintains digitally-encoded weighted associations between codewords of the source medical code and medical concepts stored in the medical-concept database and digitally-encoded weighted associations between codewords of the target medical code and medical concepts stored in the medical-concept database, the digitally-encoded weighted associations stored in one or more of the one or more electronic memories, one or more mass-storage devices, and medical-concept database.
3. The medical-code translation system of claim 2 wherein the medical-code translation system maintains digitally encoded exclusions stored in one or more of the one or more electronic memories, one or more mass-storage devices, and medical-concept database:
- code/concept exclusions;
- code/code exclusions; and
- concept/concept exclusions.
4. The medical-code translation system of claim 3 wherein the medical-code-translation system removes, from the candidate target codewords, excluded codewords to generate a set of remaining candidate target codewords by:
- removing candidate target codewords associated with a medical concept that is excluded from association with the source codeword by a code/concept exclusion;
- removing candidate target codewords excluded from association with the source codeword by a code/code exclusion; and
- removing candidate target codewords associated with a medical concept that is excluded from association with a medical concept associated with the source codeword by a concept/concept exclusion.
5. The medical-code translation system of claim 2 wherein the matching operation compares a source codeword with a target codeword by:
- computing one or more values;
- comparing each of the one or more values with a corresponding threshold value to generate a comparison value; and
- returning one or more indications of matching based on the one or more computed values and one or more comparison values.
6. The medical-code translation system of claim 5 wherein the computed values include one or more of:
- a number of medical concepts associated with the source codeword;
- a number of medical concepts associated with the target codeword;
- a number of medical concepts associated with either or both of the source codeword and the target codeword;
- a number of medical concepts associated with the source codeword that are not also associated with the target codeword;
- a number of medical concepts associated with the target codeword that are not also associated with the source codeword;
- a sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword;
- a sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword; and
- a sum of the weights of association of the medical concepts associated with the source codeword that are not also associated with the target codeword a sum of the weights of association of the medical concepts associated with the target codeword that are not also associated with the source codeword.
7. The medical-code translation system of claim 5 wherein the comparison values include one or more of:
- a value indicating whether or not a sum of the number of medical concepts associated with the source codeword that are not also associated with the target codeword and the number of medical concepts associated with the target codeword that are not also associated with the source codeword divided by the number of medical concepts associated with either or both of the source codeword and the target codeword is greater than a first threshold value;
- a value indicating whether or not a sum of the sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword and the sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword divided by the a sum of the weights of association of the medical concepts associated with the source codeword and the weights of association of the medical concepts associated with the target codeword is greater than a second threshold value; and
- a value indicating whether or not the not a sum of the sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword and the sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword divided by 2 is greater than a third threshold value.
8. The medical-code translation system of claim 5 wherein the one or more indications of matching include one or more of:
- a Boolean value indicating whether or not the source codeword matches the target codeword; and
- a numeric value indicating a degree of matching between the source codeword and the target codeword.
9. Computer instructions encoded in an electronic memory, mass-storage device, optical disk, or other physical data-storage medium that, when executed in a computer system that includes a processor, one or more electronic memories, one or more mass-storage devices, and a medical-concept database stored within, or accessible to, the computer system, implement a control program that controls operation of a medical-code-translation system that:
- receives and stores, in memory, a source codeword from a source medical code,
- identifies candidate target codewords of a target medical-concept code, using the medical-concept database, that are associated with at least one medical concept with which the source code is associated,
- removes, from the candidate target codewords, excluded codewords to generate a set of remaining candidate target codewords; and
- selects a target codeword corresponding to the source codeword that, when compared to the source codeword in a matching operation, generates a comparison metric that indicates a degree of matching with the source code greater than the degree of matching indicated by the comparison metrics generated by each of the other remaining candidate target codewords.
10. The computer instructions of claim 9 wherein the medical-code translation system maintains weighted associations between codewords of the source medical code and medical concepts stored in the medical-concept database and weighted associations between codewords of the target medical code and medical concepts stored in the medical-concept database.
11. The computer instructions of claim of 10 wherein the medical-code translation system maintains encoded exclusions, including:
- code/concept exclusions;
- code/code exclusions; and
- concept/concept exclusions.
12. The computer instructions of claim 11 wherein the medical-code-translation system removes, from the candidate target codewords, excluded codewords to generate a set of remaining candidate target codewords by:
- removing candidate target codewords associated with a medical concept that is excluded from association with the source codeword by a code/concept exclusion;
- removing candidate target codewords excluded from association with the source codeword by a code/code exclusion; and
- removing candidate target codewords associated with a medical concept that is excluded from association with a medical concept associated with the source codeword by a concept/concept exclusion.
13. The computer instructions of claim 10 wherein the matching operation compares a source codeword with a target codeword by:
- computing one or more values;
- comparing each of the one or more values with a corresponding threshold value to generate a comparison value; and
- returning one or more indications of matching based on the one or more computed values and one or more comparison values.
14. The computer instructions of claim 13 wherein the computed values include one or more of:
- a number of medical concepts associated with the source codeword;
- a number of medical concepts associated with the target codeword;
- a number of medical concepts associated with either or both of the source codeword and the target codeword;
- a number of medical concepts associated with the source codeword that are not also associated with the target codeword;
- a number of medical concepts associated with the target codeword that are not also associated with the source codeword;
- a sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword;
- a sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword; and
- a sum of the weights of association of the medical concepts associated with the source codeword and the weights of association of the medical concepts associated with the target codeword.
15. The computer instructions of claim 13 wherein the comparison values include one or more of:
- a value indicating whether or not a sum of the number of medical concepts associated with the source codeword that are not also associated with the target codeword and the number of medical concepts associated with the target codeword that are not also associated with the source codeword divided by the number of medical concepts associated with either or both of the source codeword and the target codeword is greater than a first threshold value;
- a value indicating whether or not a sum of the sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword and the sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword divided by the a sum of the weights of association of the medical concepts associated with the source codeword and the weights of association of the medical concepts associated with the target codeword is greater than a second threshold value; and
- a value indicating whether or not the not a sum of the sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword and the sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword divided by 2 is greater than a third threshold value.
16. The computer instructions of claim 13 wherein the one or more indications of matching include one or more of:
- a Boolean value indicating whether or not the source codeword matches the target codeword; and
- a numeric value indicating a degree of matching between the source codeword and the target codeword.
17. A method carried out within a computer system that includes a processor, one or more electronic memories, one or more mass-storage devices, and a medical-concept database stored within, or accessible to, the computer system, the method comprising:
- receiving and storing, in memory, a source codeword from a source medical code,
- identifying candidate target codewords of a target medical-concept code, using indications of medical concepts stored in the medical-concept database, that are associated with at least one medical concept with which the source code is associated,
- removing, from the candidate target codewords, excluded codewords to generate a set of remaining candidate target codewords; and
- selecting a target codeword corresponding to the source codeword that, when compared to the source codeword in a matching operation, generates a comparison metric that indicates a degree of matching with the source code greater than the degree of matching indicated by the comparison metrics generated for each of the other remaining candidate target codewords in matching operations.
18. The method of claim 17 further including maintaining weighted associations between codewords of the source medical code and medical concepts stored in the medical-concept database and weighted associations between codewords of the target medical code and medical concepts stored in the medical-concept database.
19. The method of claim computer 18 further including maintaining encoded exclusions, including:
- code/concept exclusions;
- code/code exclusions; and
- concept/concept exclusions.
20. The method of claim 19 wherein removing, from the candidate target codewords, excluded codewords to generate a set of remaining candidate target codewords further includes:
- removing candidate target codewords associated with a medical concept that is excluded from association with the source codeword by a code/concept exclusion;
- removing candidate target codewords excluded from association with the source codeword by a code/code exclusion; and
- removing candidate target codewords associated with a medical concept that is excluded from association with a medical concept associated with the source codeword by a concept/concept exclusion.
21. The method of claim 18 wherein the matching operation compares a source codeword with a target codeword by:
- computing one or more values;
- comparing each of the one or more values with a corresponding threshold value to generate a comparison value; and
- returning one or more indications of matching based on the one or more computed values and one or more comparison values.
22. The method of claim 21 wherein the computed values include one or more of:
- a number of medical concepts associated with the source codeword;
- a number of medical concepts associated with the target codeword;
- a number of medical concepts associated with either or both of the source codeword and the target codeword;
- a number of medical concepts associated with the source codeword that are not also associated with the target codeword;
- a number of medical concepts associated with the target codeword that are not also associated with the source codeword;
- a sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword;
- a sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword; and
- a sum of the weights of association of the medical concepts associated with the source codeword and the weights of association of the medical concepts associated with the target codeword.
23. The method of claim 21 wherein the comparison values include one or more of:
- a value indicating whether or not a sum of the number of medical concepts associated with the source codeword that are not also associated with the target codeword and the number of medical concepts associated with the target codeword that are not also associated with the source codeword divided by the number of medical concepts associated with either or both of the source codeword and the target codeword is greater than a first threshold value;
- a value indicating whether or not a sum of the sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword and the sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword divided by the a sum of the weights of association of the medical concepts associated with the source codeword and the weights of association of the medical concepts associated with the target codeword is greater than a second threshold value; and
- a value indicating whether or not the not a sum of the sum of the weights of association of the medical concepts associated with the source codeword that are also associated with the target codeword and the sum of the weights of association of the medical concepts associated with the target codeword that are also associated with the source codeword divided by 2 is greater than a third threshold value.
24. The method of claim 21 wherein the one or more indications of matching include one or more of:
- a Boolean value indicating whether or not the source codeword matches the target codeword; and
- a numeric value indicating a degree of matching between the source codeword and the target codeword.
Type: Application
Filed: May 16, 2012
Publication Date: Nov 21, 2013
Inventors: Parag Patel (Bellevue, WA), Abhishek Jacob (Bellevue, WA), Virendra Prasad (Bellevue, WA), Vijay Bhuttar (Bellevue, WA), Ryan McDermitt (Bellevue, WA)
Application Number: 13/472,767
International Classification: G06F 17/30 (20060101);