Determining a next-best action across medical conditions
A computational evidence platform extracts clinical concepts from medical evidence sources and creates a database of elemental diagnostic factors and elemental investigations links to medical conditions. Input from a person groups factors and investigations makes corrections and adds a ranking Elemental factors and investigations do not include information specific to their associated conditions but include synonyms and a link to a medical ontology. A patient state is determined by extracting patient known diagnostic factors and investigation results from the patient chart. These known factors and results are matched to the database and a ranking of likely conditions are output. Next-best actions per condition are output by determining factors not yet known and investigations not yet performed. Next-best actions across conditions are determined by performing a recursive tree search of the database and assuming that unknown factors are now known to generate a score for each assumption.
1. A method of determining a differential diagnosis for a patient, said method comprising:
receiving, at a diagnostic software application, health information of a patient from a computer system during a patient encounter or when said patient is not present, wherein said diagnostic software application includes a user interface;
automatically extracting patient diagnostic factors known to be present or absent in said patient from said received health information;
automatically matching said patient diagnostic factors to diagnostic factor nodes in a knowledge base, said diagnostic factor nodes being linked to medical condition nodes in said knowledge base;
automatically ranking medical conditions that are each associated with one of said medical condition nodes based upon said diagnostic factor nodes;
displaying on a computing device said ranked medical conditions;
assuming that a diagnostic factor of said patient not known to be present or absent is now known to be present or absent, that a laboratory test of said patient not performed has now been performed or that an examination of said patient not performed has now been performed;
automatically generating a new ranking of said medical conditions based upon said assuming, and determining at least one next-best action on the basis of said new ranking of said medical conditions; and
displaying said next-best action on said computing device.
2. A method as recited in claim 1 wherein said next-best action is said diagnostic factor of said patient not yet known to be present or absent, said laboratory test of said patient not yet performed or said examination of said patient not yet performed.
3. A method as recited in claim 1 further comprising:
displaying on said computing device said ranked medical conditions in order of rank.
4. A method as recited in claim 1 wherein said displaying said ranked medical conditions occurs before a patient encounter with a clinician.
5. A method as recited in claim 1 wherein said extracting further comprises:
extracting patient diagnostic factors known to be present or absent in said patient from an electronic health record of said patient.
6. A method as recited in claim 1 wherein said diagnostic software application executes upon a cloud-based computer, and wherein said displaying said ranked medical conditions occurs within a Web browser of said computing device.
7. A method as recited in claim 1 wherein each of said rankings indicates a probability that one of said medical conditions is due to said patient diagnostic factors.
8. A method as recited in claim 1 wherein said health information is received from said computer system of said patient via an online chat, via an online session, or via a video call.
9. A method as recited in claim 1 wherein said health information is received from an electronic health record (EHR) of said computer system or from said computer system via a query API (application programming interface) of said knowledge base.
10. A method of determining a differential diagnosis for a patient, said method comprising:
receiving, at a diagnostic software application, health information of a patient from a computer system during a patient encounter or when said patient is not present, wherein said diagnostic software application includes a user interface;
automatically extracting patient diagnostic factors known to be present or absent in said patient from said received health information;
automatically matching said patient diagnostic factors to diagnostic factor nodes in a knowledge base, said diagnostic factor nodes being linked to medical condition nodes in said knowledge base;
automatically ranking medical conditions that are each associated with one of said medical condition nodes based upon said diagnostic factor nodes;
displaying on a computing device said ranked medical conditions;
automatically ranking said medical conditions again by assuming that an unknown diagnostic factor of said patient currently not known to be present or absent is now known to be present or absent, that a laboratory test of said patient not performed has now been performed or that an examination of said patient not performed has now been performed in order to obtain newly-ranked medical conditions;
automatically determining that a next-best action is said unknown diagnostic factor, said laboratory test or said examination of said patient based upon said newly-ranked medical conditions; and
displaying said next-best action on said computing device.
11. A method as recited in claim 10 wherein said health information is received from said computer system of said patient via an online chat, via an online session, or via a video call.
12. A method as recited in claim 10 wherein said health information is received from an electronic health record (EHR) of said computer system or from said computer system via a query API (application programming interface) of said knowledge base.
13. A method as recited in claim 10 wherein said diagnostic software application executes upon a cloud-based computer, and wherein said displaying said ranked medical conditions occurs within a Web browser of said computing device.
14. A method of outputting a next-best action for a patient, said method comprising:
receiving, at a knowledge graph, patient diagnostic factors known to be present or absent in said patient;
automatically matching said patient diagnostic factors to diagnostic factor nodes in said knowledge graph, said diagnostic factor nodes being linked to medical condition nodes in said knowledge graph, each medical condition node indicating a medical condition;
automatically ranking said medical conditions by assuming that an unknown diagnostic factor of said patient currently not known to be present or absent is now known to be present or absent, in order to obtain ranked medical conditions;
automatically determining that a next-best action is said unknown diagnostic factor of said patient based upon said ranked medical conditions; and
outputting said next-best action to a diagnostic software application of a computing device during a patient encounter or when said patient is not present, wherein said diagnostic software application includes a user interface.
15. A method as recited in claim 14 further comprising:
generating a score for said unknown diagnostic factor that represents a change in a differential diagnosis between said ranked medical conditions when said unknown diagnostic factor is now known to be present or absent; and
determining that said next-best action is said unknown diagnostic factor based upon said score.
16. A method as recited in claim 14 further comprising:
ranking said medical conditions by assuming that said unknown diagnostic factor of said patient currently not known to be present or absent is now known to be present or absent, that a laboratory test of said patient not performed has now been performed or that an examination of said patient not performed has now been performed in order to obtain said ranked medical conditions; and
determining that said next-best action is said unknown diagnostic factor of said patient, said laboratory test or said examination of said patient based upon said ranked medical conditions.
17. A method as recited in claim 14 further comprising:
outputting said next-best action from said knowledge graph to said computing device via an application programming interface (API).
18. A method as recited in claim 14 further comprising:
performing a search algorithm that assumes that a plurality of unknown diagnostic factors of said patient each currently not known to be present or absent are each now known to be present or absent, in order to obtain a score for said each unknown diagnostic factor;
determining a plurality of next-best actions based upon said scores, each of said next-best actions being one of said unknown diagnostic factors of said patient assumed to be known present or absent; and
outputting said next-best actions to a computing device.
19. A method as recited in claim 14 wherein said patient diagnostic factors are received from a computer of a patient via an online chat, via an online session, or via a video call.
20. A method as recited in claim 14 wherein said patient diagnostic factors are received from an electronic health record (EHR) of a computer system or from a computer system via a query API (application programming interface) of said knowledge graph.