Valence profiling of virtual interactive objects
Introduced here are health management platforms able to infer the health state of a subject based on behavioral changes reflected in the digital activities performed by a subject. A health management platform can initially acquire contextual data pertaining to digital activities performed by the subject. The health management platform can identify the target(s) of the digital activities by parsing the contextual data, and then compile the digital activities and corresponding activities into a personalized valence index. Thereafter, the health management platform can compare digital activities performed by the subject to the entries included in the personalized valence index. If a matching entry is not discovered, the health management platform can identify an instance of behavior change. These instances of behavior change may be indicative of changes in the subject's health state.
1 . A non-transitory medium with instructions stored thereon that, when executed by a processor of a computing device, cause the computing device to perform operations comprising:
acquiring, in real time, contextual data that pertains to digital activities performed by an individual using a computer program executing on the computing device;
generating electronic records of the digital activities by associating each digital activity with a target that is identified via analysis of the contextual data,
wherein each target is representative of a person or a geographical location that is involved in a corresponding one of the digital activities;
calculating valence measures for the digital activities based on the targets identified for the digital activities;
configuring a profile that is associated with the individual to include associations between the valence measures and the electronic records of the digital activities;
applying a personalized algorithm to the profile to detect changes in health of the individual at a temporal resolution that exceeds clinical interactions based on dissimilarities between valence measures calculated for digital activities and preceding digital activities, wherein the personalized algorithm is configured for the individual and is continuously trained in real time based on the valence measures and the electronic records of the digital activities;
establishing, based on an output produced by the personalized algorithm upon being applied to the profile, that there has been a change in health of the individual in response to a determination that a given valence measure calculated for a given digital activity indicates at least a given degree of dissimilarity from one or more valence measures of one or more preceding digital activities; and
continuously updating the calculated valence measures using the personalized algorithm to constantly establish change in health of the individual in real time.
2 . The non-transitory medium of claim 1 , wherein said generating, said calculating, said configuring, and said establishing are performed in real time, so as to promptly detect changes in the health of the individual.
3 . The non-transitory medium of claim 1 , wherein the operations further comprise:
causing, in response to said establishing, presentation of a notification that recommends an additional action to address the change in the health of the individual.
4 . The non-transitory medium of claim 3 , wherein the additional action involves performance of one of the digital activities for which the corresponding valence measure is positive.
5 . The non-transitory medium of claim 1 , wherein each of the valence measures is based on (i) paralinguistic analysis, (ii) non-linguistic analysis, or (iii) linguistic analysis of a corresponding one of the digital activities.
6 . The non-transitory medium of claim 1 , wherein each of the valence measures characterizes how the individual feels emotionally about the target that is identified via analysis of the contextual data.
7 . The non-transitory medium of claim 1 , wherein the computer program is a social media application.
8 . The non-transitory medium of claim 7 , wherein the operations further comprise:
establishing communication with a social media platform that is responsible for managing the social media application via an application programming interface;
wherein the contextual data is acquired via the application programming interface.
9 . A method for documenting digital activities of an individual in order to discover changes in health, the method comprising:
acquiring contextual data that is associated with the individual and that pertains to digital activities performed with, or monitored by, a computer program executing on a computing device;
in response to said acquiring,
generating electronic records of the digital activities by associating each digital activity with a target that is representative of a person or a geographical location that is involved in a corresponding one of the digital activities;
calculating valence measures for the digital activities based on the targets identified for the digital activities; and
configuring a profile that is associated with the individual to include associations between the valence measures and the electronic records of the digital activities; and
applying a machine learning model to the profile at a temporal resolution that exceeds clinical interactions to detect dissimilarities between valence measures calculated for digital activities and preceding digital activities, wherein the machine learning model is trained in a personalized manner using a dataset that includes (i) past digital activities performed with, or monitored by, the computer program executing on the computing device that is known to be associated with an individual and (ii) accompanying labels, each of which is provided by the individual and specifies a valence measure for a corresponding one of the past digital activities, and wherein the machine learning model is updated in real time based on the valence measures and the electronic records of the digital activities;
establishing, based on an output produced by the machine learning model upon being applied to the profile, that there has been a change in health of the individual in response to a determination that a given valence measure calculated for a given digital activity indicates at least a given degree of dissimilarity from one or more valence measures of one or more preceding digital activities; and
continuously updating the calculated valence measures using the personalized machine learning model to constantly establish change in health of the individual in real time.
10 . The method of claim 9 , wherein the digital activities include movement activities, and wherein for each of the digital activities, the target is a geographical location.
11 . The method of claim 9 , wherein the digital activities include engagements with a social media application, and wherein for each of the digital activities, the target is a social media identifier that is representative of a person.
12 . The method of claim 11 , wherein the social media identifier is an account identifier or a communication identifier.
13 . The method of claim 9 , wherein each of the electronic records further specifies whether the individual initiates a corresponding one of the digital activities.
14 . The method of claim 9 , further comprising:
applying, to the profile, a machine learning model that is trained to learn relationships between past digital activities and accompanying labels that are representative of valence measures.
15 . The method of claim 14 , wherein the past digital activities involve one or more individuals that share a characteristic in common with the individual.
16 . The method of claim 15 , wherein the characteristic is age, sex, or disease classification.
17 . A method for documenting digital activities of an individual in order to discover changes in health, the method comprising:
acquiring contextual data that is associated with the individual and that pertains to digital activities performed with, or monitored by, a computer program executing on a computing device;
in response to said acquiring,
generating electronic records of the digital activities by associating each digital activity with a target that is representative of a person or a geographical location that is involved in a corresponding one of the digital activities;
calculating valence measures for the digital activities based on the targets identified for the digital activities; and
configuring a profile that is associated with the individual to include associations between the valence measures and the electronic records of the digital activities; and
applying a machine learning model to the profile at a temporal resolution that exceeds clinical interactions to detect dissimilarities between valence measures calculated for digital activities and preceding digital activities, wherein the machine learning model is fine-tuned using a dataset that includes past digital activities performed with, or monitored by, the computer program executing on the computing device that is known to be associated with an individual; and
establishing, based on an output produced by the machine learning model upon being applied to the profile, that there has been a change in health of the individual in response to a determination that a given valence measure calculated for a given digital activity indicates at least a given degree of dissimilarity from one or more valence measures of one or more preceding digital activities; and
continuously updating the calculated valence measures using the fine-tuned machine learning model to constantly establish change in health of the individual in real time.