Methods and apparatus for providing guidance to medical professionals

1. A system comprising:

at least one processor; and

at least one storage medium storing executable instructions that, when executed by the at least one processor, cause the at least one processor to carry out a method comprising:

evaluating a medical recommendation of a medical professional with respect to a patient encounter to determine whether the medical recommendation is consistent with at least one medical guideline, wherein evaluating the medical recommendation to determine whether the medical recommendation complies with the at least one medical guideline comprises:

extracting, using a natural language understanding engine and in real-time, one or more facts from at least one narrative provided by the medical professional in connection with the patient encounter, wherein at least one fact of the one or more extracted facts is a medical finding, and wherein the natural language understanding engine uses a machine learning technique selected from the group consisting of maximum entropy modeling, support vector machines, and random fields, wherein the machine learning technique is trained on a training corpus of hand-annotated medical documentation that was previously hand-annotated by a human labeler with expert knowledge of the domain to identify all extracted features and associated entity labels describing facts in the documentation, the training including learning, for each extracted feature in the training corpus, a probability with which tokens having that feature are associated with each entity type, and wherein the training corpus of hand-annotated medical documentation is utilized to train the machine learning technique to automatically label input tokens of the at least one narrative provided by the medical professional based on the learned probability;

selecting, based on at least the medical finding and from among a set of medical guidelines, the at least one medical guideline that is applicable to the patient encounter; and

determining, using the one or more facts extracted from the at least one narrative, whether the medical recommendation is consistent with the at least one medical guideline, wherein determining whether the medical recommendation complies with the at least one medical guideline comprises:

applying one or more decision rules associated with the at least one medical guideline to the one or more facts extracted from the at least one narrative, including the medical finding, the one or more decision rules comprising at least one recommendation called for by the at least one medical guideline;

determining whether the medical recommendation is consistent with the at least one recommendation called for by the at least one medical guideline; and

generating an audible alert when the medical recommendation is inconsistent with the at least one recommendation called for by the at least one medical guideline, wherein the alert indicates the at least one recommendation called for by the at least one medical guideline, and wherein the at least one recommendation called for by the at least one medical guideline is accepted or rejected by the medical professional.

2. The system of claim 1 , wherein:

the one or more decision rules are associated with one or more items of information called for by the at least one medical guideline and a plurality of values and/or one or more synonyms for one or more of the plurality of values for the one or more items of information, and

determining whether the medical recommendation of the medical professional complies with the at least one medical guideline comprises:

determining whether the medical professional provided in the at least one narrative the one or more items of information called for by the at least one medical guideline, wherein determining whether the medical professional provided in the at least one narrative the one or more items of information comprises determining whether the at least one narrative includes any of the values or the synonyms for the one or more items of information.

3. The system of claim 2 , wherein the method further comprises:

in response to determining that the medical professional did not provide in the at least one narrative at least a first item of information, of the one or more items of information called for by at least the first medical guideline:

automatically selecting an assumed value from a plurality of possible values to use for the first item of information that was not provided in the at least one narrative; and

applying the one or more decision rules to the one or more facts and the assumed value to determine whether the medical recommendation is consistent with the at least one recommendation called for by at least one medical guideline.

4. The system of claim 3 , wherein the patient encounter is associated with a particular patient, and wherein selecting the assumed value comprises selecting based at least in part on medical history associated with the particular patient and/or statistical information measured from a population relevant for the particular patient.

5. The system of claim 3 , wherein the method further comprises:

receiving from the at least one medical professional a correction to the assumed value for the first item of information that was not included in the at least one narrative; and

reapplying at least one of the one or more decision rules based at least in part on the correction received from the at least one medical professional.

6. The system of claim 1 , wherein the method further comprises:

generating the one or more decision rules based at least in part on the at least one medical guideline, wherein generating the one or more decision rules comprises:

identifying, from the at least one medical guideline, the at least one recommendation called for by the at least one medical guideline; and

generating, at least a first decision rule of the one or more decision rules, that includes the at least one recommendation called for by the at least one medical guideline.

7. The system of claim 6 , wherein the method further comprises:

monitoring one or more sources from which published guidelines are available, wherein monitoring the one or more sources comprises detecting that the at least one medical guideline is newly available from one of the one or more sources; and

generating the one or more decision rules comprises generating at least one new decision rule based on the at least one medical guideline that is newly available.

8. A method comprising acts of:

evaluating a medical recommendation of a medical professional with respect to a patient encounter to determine whether the medical recommendation is consistent with at least one medical guideline, wherein evaluating the medical recommendation to determine whether the medical recommendation complies with the at least one medical guideline comprises:

extracting, using a natural language understanding engine and in real-time, one or more facts from at least one narrative provided by the medical professional in connection with the patient encounter, wherein the natural language understanding engine uses a machine learning technique selected from the group consisting of maximum entropy modeling, support vector machines, and random fields, wherein the machine learning technique is trained on a training corpus of hand-annotated medical documentation that was previously hand-annotated by a human labeler with expert knowledge of the domain to identify all extracted features and associated entity labels describing facts in the documentation, the training including learning, for each extracted feature in the training corpus, a probability with which tokens having that feature are associated with each entity type in the training corpus of hand-annotated medical documentation is utilized to train the machine learning technique to automatically label input tokens of the at least one narrative provided by the medical professional based on the learned probability;

selecting, based on the one or more facts extracted from the at least one narrative and from among a set of medical guidelines, the at least one medical guideline that is applicable to the patient encounter; and

determining, using the one or more facts extracted from the at least one narrative, whether the medical recommendation is consistent with the at least one medical guideline, wherein determining whether the medical recommendation complies with the at least one medical guideline comprises:

applying one or more decision rules associated with the at least one medical guideline to the one or more facts extracted from the at least one narrative, the one or more decision rules comprising at least one recommendation called for by the at least one medical guideline;

determining whether the medical recommendation is consistent with the at least one recommendation called for by the at least one medical guideline; and

generating an audible alert when the medical recommendation is inconsistent with the at least one recommendation called for by the at least one medical guideline, wherein the alert indicates the at least one recommendation called for by the at least one medical guideline, and wherein the at least one recommendation called for by the at least one medical guideline is accepted or rejected by the medical professional.

9. The method of claim 8 , wherein:

the one or more decision rules are associated with one or more items of information called for by the at least one medical guideline and a plurality of values and/or one or more synonyms for one or more of the plurality of values for the one or more items of information, and

determining whether the medical recommendation of the medical professional complies with the at least one medical guideline comprises:

determining whether the medical professional provided in the at least one narrative the one or more items of information called for by the at least one medical guideline, wherein determining whether the medical professional provided in the at least one narrative the one or more items of information comprises determining whether the at least one narrative includes any of the values or the synonyms for the one or more items of information.

10. The method of claim 9 , wherein the one or more decision rules further comprise at least one statement corresponding to at least one decision to be made with respect to the one or more items of information and the at least one medical guideline.

11. The method of claim 9 , further comprising:

in response to determining that the medical professional did not provide in the at least one narrative at least a first item of information, of the one or more items of information called for by at least the first medical guideline:

automatically selecting an assumed value from a plurality of possible values to use for the first item of information that was not provided in the at least one narrative; and applying the one or more decision rules to the one or more facts and the assumed value to determine whether the medical recommendation is consistent with the at least one recommendation called for by at least one medical guideline.

12. The method of claim 11 , wherein the patient encounter is associated with a particular patient, and wherein selecting the assumed value comprises selecting based at least in part on medical history associated with the particular patient and/or statistical information measured from a population relevant for the particular patient.

13. The method of claim 11 , further comprising:

receiving from the at least one medical professional a correction to the assumed value for the first item of information that was not included in the at least one narrative; and

reapplying at least one of the one or more decision rules based at least in part on the correction received from the at least one medical professional.

14. The method of claim 8 , further comprising:

generating the one or more decision rules based at least in part on the at least one medical guideline, wherein generating the one or more decision rules comprises:

identifying, from the at least one medical guideline, the at least one recommendation called for by the at least one medical guideline; and

generating, at least a first decision rule of the one or more decision rules, that includes the at least one recommendation called for by the at least one medical guideline.

15. The method of claim 14 , further comprising:

monitoring one or more sources from which published guidelines are available, wherein monitoring the one or more sources comprises detecting that the at least one medical guideline is newly available from one of the one or more sources; and

generating the one or more decision rules comprises generating at least one new decision rule based on the at least one medical guideline that is newly available.

16. The method of claim 8 , wherein determining whether the medical recommendation is consistent with the at least one recommendation called for by the at least one medical guideline comprises determining whether the medical professional did not provide, in the at least one narrative, a recommendation identified by the at least one medical guideline.

17. At least one non-transitory computer-readable storage medium having stored thereon instructions that, when executed by at least one processor, perform a method comprising acts of:

evaluating a medical recommendation of a medical professional with respect to a patient encounter to determine whether the medical recommendation is consistent with at least one medical guideline, wherein evaluating the medical recommendation to determine whether the medical recommendation complies with the at least one medical guideline comprises:

extracting, using a natural language understanding engine and in real-time, one or more facts from at least one narrative provided by the medical professional in connection with the patient encounter, wherein the natural language understanding engine uses a machine learning technique selected from the group consisting of maximum entropy modeling, support vector machines, and random fields, wherein the machine learning technique is trained on a training corpus of hand-annotated medical documentation that was previously hand-annotated by a human labeler with expert knowledge of the domain to identify all extracted features and associated entity labels describing facts in the documentation, the training including learning, for each extracted feature in the training corpus, a probability with which tokens having that feature are associated with each entity type in the training corpus of hand-annotated medical documentation is utilized to train the machine learning technique to automatically label input tokens of the at least one narrative provided by the medical professional based on the learned probability;

selecting, based on the one or more facts extracted from the at least one narrative and from among a set of medical guidelines, the at least one medical guideline that is applicable to the patient encounter; and

determining, using the one or more facts extracted from the at least one narrative, whether the medical recommendation is consistent with the at least one medical guideline, wherein determining whether the medical recommendation complies with the at least one medical guideline comprises:

applying one or more decision rules associated with the at least one medical guideline to the one or more facts extracted from the at least one narrative, the one or more decision rules comprising at least one recommendation called for by the at least one medical guideline;

determining whether the medical recommendation is consistent with the at least one recommendation called for by the at least one medical guideline; and

generating an audible alert when the medical recommendation is inconsistent with the at least one recommendation called for by the at least one medical guideline, wherein the alert indicates the at least one recommendation called for by the at least one medical guideline, and wherein the at least one recommendation called for by the at least one medical guideline is accepted or rejected by the medical professional.