Intelligent computer application for diagnosis suggestion and validation
In some examples, data associated one or more medical presentations of a person is received. A mapping that associates the data to one or more candidate diagnoses is determined. The one or more candidate diagnoses correspond to one or more possible medical conditions associated with the person. A validation score for a first candidate diagnosis, of the one or more candidate diagnoses, is generated based on a set of features associated with the person. The validation score indicates an accuracy of the candidate diagnosis. At least partially in response to the generating of the validation score, an indicator is caused to be provided at a user device. The indicator indicates the validation score.
1 . A computer-implemented method performed by one or more hardware processors, the computer-implemented method comprising:
determining a mapping that associates a first set of data to one or more candidate diagnoses based on one or more predetermined rules and based further on a first set of features, wherein the first set of data is associated with one or more medical presentations of a person, and wherein the one or more candidate diagnoses correspond to one or more possible medical conditions associated with the person;
generating a validation score for a first candidate diagnosis of the one or more candidate diagnoses based on a second set of features, associated with the person, and based further on one or more machine learning electronic models,
wherein:
(a) the one or more machine learning electronic models are trained by inputting, to the one or more machine learning electronic models, vectors of information corresponding to instances of the second set of features associated with people, wherein training the one or more machine learning electronic models is based at least in part on: (i) learning parameters or learning weights, the learning parameters or learning weights associated with the second set of features, and (ii) converting or encoding inputs into corresponding feature vectors,
(b) the generating comprises applying, to the trained one or more machine learning electronic models, a second set of data associated with the second set of features to facilitate the generation of the validation score, the second set of features differing from the first set of features and the validation score indicating an accuracy of the first candidate diagnosis, and
(c) the validation score indicates an accuracy of the first candidate diagnosis; and
in response to the generating, causing an indicator to be generated at an electronic graphical user interface, wherein the indicator is displayed on the electronic graphical user interface and conveys information including the validation score.
2 . The computer-implemented method of claim 1 , wherein the first set of features comprises feature data selected from a group comprising: data that specifies a chief complaint of the person, data that specifies a reason for a visit by the person to a healthcare practitioner, and data presenting problems of the person.
3 . The computer-implemented method of claim 1 , wherein the information identifies that the first candidate diagnosis is accurate, the first candidate diagnosis is moderately accurate, or the first candidate diagnosis is minimally accurate.
4 . The computer-implemented method of claim 3 , wherein the first set of features indicates a feature selected from a group of attributes comprising age of the person, gender of the person, race of the person, procedure history of the person, and the first set of data, and wherein a first attribute value of a first attribute associated with the group of attributes is determined based on a second attribute value of a second attribute associated with the group of attributes.
5 . The computer-implemented method of claim 1 , wherein the second set of features but not the first set of features indicates a feature selected from a group comprising patient sensor data, user physiological data, and social-media data of the person.
6 . The computer-implemented method of claim 1 , wherein the second set of features but not the first set of features comprises one or both of patient sensor data and user device physiological data.
7 . The computer-implemented method of claim 1 , wherein the validation score indicates that: the first candidate diagnosis is maximally accurate over a first threshold, the first candidate diagnosis is moderately accurate between a second threshold and the first threshold, or the first candidate diagnosis is minimally accurate under the second threshold.
8 . The computer-implemented method of claim 1 , wherein the indicator corresponds to a presentation element selected from a group comprising a maximally accurate validation score, a moderately accurate validation score, and a minimally accurate validation score.
9 . One or more non-transitory media having computer-readable instructions that, when executed by one or more hardware processors, cause the one or more hardware processors to perform a plurality of operations, the operations comprising:
determining a mapping that associates a first set of data to one or more candidate diagnoses based on one or more predetermined rules and based further on a first set of features, wherein the first set of data is associated with one or more medical presentations of a person, and wherein the one or more candidate diagnoses correspond to one or more possible medical conditions associated with the person;
generating a validation score for a first candidate diagnosis of the one or more candidate diagnoses based on a second set of features, associated with the person, and based further on one or more machine learning electronic models,
wherein:
(a) the one or more machine learning electronic models are trained by inputting, to the one or more machine learning electronic models, vectors of information corresponding to instances of the second set of features associated with people, wherein training the one or more machine learning electronic models is based at least in part on: (i) learning parameters or learning weights, the learning parameters or learning weights associated with the second set of features, and (ii) converting or encoding inputs into corresponding feature vectors,
(b) the generating comprises applying, to the trained one or more machine learning electronic models, a second set of data associated with the second set of features to facilitate the generation of the validation score, the second set of features differing from the first set of features and the validation score indicating an accuracy of the first candidate diagnosis, and
(c) the validation score indicates an accuracy of the first candidate diagnosis; and
in response to the generating, causing an indicator to be generated at an electronic graphical user interface, wherein the indicator is displayed on the electronic graphical user interface and conveys information including the validation score.
10 . The one or more non-transitory media of claim 9 , wherein the first set of features comprises feature data selected from a group comprising: data that specifies a chief complaint of the person, data that specifies a reason for a visit by the person to a healthcare practitioner, and data presenting problems of the person.
11 . The one or more non-transitory media of claim 9 , wherein the information identifies that the first candidate diagnosis is accurate, the first candidate diagnosis is moderately accurate, or the first candidate diagnosis is minimally accurate.
12 . The one or more non-transitory media of claim 9 , wherein the second set of features but not the first set of features indicates a feature selected from a group comprising patient sensor data, user physiological data, and social-media data of the person.
13 . The one or more non-transitory media of claim 9 , wherein the second set of features but not the first set of features comprises one or both of patient sensor data and user device physiological data.
14 . The one or more non-transitory media of claim 9 , wherein the validation score indicates that: the first candidate diagnosis is maximally accurate over a first threshold, the first candidate diagnosis is moderately accurate between a second threshold and the first threshold, or the first candidate diagnosis is minimally accurate under the second threshold.
15 . The one or more non-transitory media of claim 9 , wherein the information corresponds to an element selected from a group comprising a first item associated with a first validation score, a second item associated with a second validation score and with a second accuracy less than a first accuracy associated with the first validation score, and a third item associated with a third validation score and a third accuracy less than the second accuracy.
16 . A system having one or more hardware processors configured to perform a plurality of operations, the operations comprising:
determining a mapping that associates a first set of data to one or more candidate diagnoses based on one or more predetermined rules and based further on a first set of features, wherein the first set of data is associated with one or more medical presentations of a person, and wherein the one or more candidate diagnoses correspond to one or more possible medical conditions associated with the person;
generating a validation score for a first candidate diagnosis of the one or more candidate diagnoses based on a second set of features, associated with the person, and based further on one or more machine learning electronic models,
wherein:
(a) the one or more machine learning electronic models are trained by inputting, to the one or more machine learning electronic models, vectors of information corresponding to instances of the second set of features associated with people, wherein training the one or more machine learning electronic models is based at least in part on: (i) learning parameters or learning weights, the learning parameters or learning weights associated with the second set of features, and (ii) converting or encoding inputs into corresponding feature vectors,
(b) the generating comprises applying, to the trained one or more machine learning electronic models, a second set of data associated with the second set of features to facilitate the generation of the validation score, the second set of features differing from the first set of features and the validation score indicating an accuracy of the first candidate diagnosis, and
(c) the validation score indicates an accuracy of the first candidate diagnosis; and
in response to the generating, causing an indicator to be generated at an electronic graphical user interface, wherein the indicator is displayed on the electronic graphical user interface and conveys information including the validation score.
17 . The system of claim 16 , wherein the validation score indicates that: the first candidate diagnosis is maximally accurate over a first threshold, the first candidate diagnosis is moderately accurate between a second threshold and the first threshold, or the first candidate diagnosis is minimally accurate under the second threshold.
18 . The system of claim 16 , wherein the second set of features but not the first set of features comprises one or both of patient sensor data and user device physiological data.
19 . The system of claim 16 , wherein the information identifies a first accuracy level, a second accuracy level differing from the first accuracy level, or a third accuracy level differing from the first accuracy level and from the second accuracy level.
20 . The system of claim 19 , wherein the first accuracy level and the second accuracy level are greater than the second accuracy level and the third accuracy level, respectively.
21 . The system of claim 17 , wherein the one or more hardware processors are further configured to perform the determining, the generating, and the causing automatically without user input.