User Behavior Monitoring On A Computerized Device

Abstract

The subject disclosure is directed towards monitoring user behavior on a computerized device for any deviation from normal or acceptable behavior that is likely to affect the user's mental state. A prediction model corresponding to features of one or more mental states may be compared with features based upon current user behavior. If the user's current behavior presents a mental state indicative of an uncharacteristic deviation from the normal or acceptable behavior, descriptive data associated with that mental state may be presented to the user in addition to a trusted individual, such as a health care professional.

1 .- 20 . (canceled) 
     
     
         21 . A method performed at least in part on at least one processor comprising:
 accessing, by a training mechanism implemented on the at least one processor, a decision tree corresponding to at least one mental state;   comparing the decision tree with an expert assessment corresponding to the at least one mental state;   determining whether to update the decision tree based on the comparison of the decision tree and the expert assessment;   responsive to a determination to update the decision tree, modifying the decision tree; and   generating a classifier for predicting psychological events from user interaction data comprising one or more of search engine queries and social media postings based on the modified decision tree corresponding to the at least one mental state.   
     
     
         22 . The method of  claim 21 , wherein the decision tree is configured to identify behavior changes indicative of an effect on the at least one mental state. 
     
     
         23 . The method of  claim 21 , wherein the decision tree is a prediction model. 
     
     
         24 . The method of  claim 21 , wherein comparing the decision tree with the expert assessment further comprises:
 receiving at least one of an annotation, addition, removal, condition, or modification to the decision tree as the expert assessment corresponding to the at least one mental state.   
     
     
         25 . The method of  claim 24 , wherein determining whether to update the decision tree based on the comparison further comprises:
 analyzing the received expert assessment to identify a number of errors associated with the decision tree;   determining whether the identified number of errors exceeds a threshold;   responsive to a determination that the identified number of errors does not exceed the threshold, generating the classifier; and   responsive to a determination that the identified number of errors exceeds the threshold, modifying the decision tree.   
     
     
         26 . The method of  claim 24 , wherein determining whether to update the decision tree based on the comparison further comprises:
 analyzing the received expert assessment to identify a measure associated with an increase in a detection success rate for the at least one mental state;   determining whether the identified measure exceeds a threshold;   responsive to a determination that the identified measure does not exceed the threshold, generating the classifier; and   responsive to a determination that the identified measure exceeds the threshold, modifying the decision tree.   
     
     
         27 . The method of  claim 21 , wherein modifying the decision tree further comprises:
 modifying at least one comparison operation based on the comparison between the decision tree and the expert assessment.   
     
     
         28 . In a computing environment, a computerized device, comprising:
 a prediction model corresponding to at least one mental state; and   a training mechanism configured to:
 access the prediction model corresponding to the at least one mental state; 
 compare the prediction model with an expert assessment corresponding to the at least one mental state; 
 determine whether to update the decision tree based on the comparison of the decision tree and the expert assessment; 
 modify the decision tree in response to a determination to update the decision tree; and 
 generate a classifier for predicting psychological events based on the modified decision tree. 
   
     
     
         29 . The computerized device or the configuration of coupled computerized devices of  claim 28 , wherein the training mechanism is further configured to:
 receive at least one of an annotation, addition, removal, condition, or modification to the decision tree as the expert assessment corresponding to the at least one mental state.   
     
     
         30 . The computerized device or the configuration of coupled computerized devices of  claim 29 , wherein the training mechanism is further configured to:
 analyze the received expert assessment to identify a number of errors associated with the decision tree;   determine whether the identified number of errors exceeds a threshold;   responsive to a determination that the identified number of errors does not exceed the threshold, proceed to the generating step; and   responsive to a determination that the identified number of errors exceeds the threshold, proceed to the modifying step.   
     
     
         31 . The computerized device or the configuration of coupled computerized devices of  claim 28 , wherein the training mechanism is further configured to:
 modify at least one comparison operation based on the comparison between the decision tree and the expert assessment.   
     
     
         32 . The computerized device or the configuration of coupled computerized devices of  claim 28 , wherein the generated classifier is downloadable from a server for use by a monitoring component implemented on a computing device. 
     
     
         33 . The computerized device or the configuration of coupled computerized devices of  claim 28 , wherein the generated classifier is configured to operate on a server as a network service for a monitoring component implemented on a computing device. 
     
     
         34 . The computerized device or the configuration of coupled computerized devices of  claim 28 , wherein the training mechanism is further configured to:
 receive user feedback corresponding to the generated classifier for the at least one mental state; and   continuously modify the prediction model for the at least one mental state based on the received user feedback.   
     
     
         35 . The computerized device or the configuration of coupled computerized devices of  claim 28 , wherein the training mechanism is further configured to:
 determine whether an amount of additional training data available corresponding to the at least one mental state exceeds a threshold; and   responsive to a determination that the amount exceeds the threshold, rebuilding a new instance of the prediction model for the at least one mental state using the additional training data and existing training data.   
     
     
         36 . One or more computer storage media having computer-executable instructions, which on execution by a computer cause the computer to perform operations, comprising:
 accessing, by a training mechanism implemented on the at least one processor, a decision tree corresponding to at least one mental state;   comparing the decision tree with an expert assessment corresponding to the at least one mental state;   determining whether to update the decision tree based on the comparison of the decision tree and the expert assessment;   responsive to a determination to update the decision tree, modifying the decision tree; and   generating a classifier for predicting psychological events based on the modified decision tree.   
     
     
         37 . The one or more computer storage media of  claim 36 , wherein comparing the decision tree with the expert assessment further comprises:
 receiving at least one of an annotation, addition, removal, condition, or modification to the decision tree as the expert assessment corresponding to the at least one mental state.   
     
     
         38 . The one or more computer storage media of  claim 37 , wherein determining whether to update the decision tree based on the comparison further comprises:
 analyzing the received expert assessment to identify a number of errors associated with the decision tree;   determining whether the identified number of errors exceeds a threshold;   responsive to a determination that the identified number of errors does not exceed the threshold, proceeding to the generating step; and   responsive to a determination that the identified number of errors exceeds the threshold, proceeding to the modifying step.   
     
     
         39 . The one or more computer storage media of  claim 37 , wherein determining whether to update the decision tree based on the comparison further comprises:
 analyzing the received expert assessment to identify a measure associated with an increase in a detection success rate for the at least one mental state;   determining whether the identified measure exceeds a threshold;   responsive to a determination that the identified measure does not exceed the threshold, proceeding to the generating step; and   responsive to a determination that the identified measure exceeds the threshold, proceeding to the modifying step.   
     
     
         40 . The one or more computer storage media of  claim 36 , wherein modifying the decision tree further comprises:
 modifying at least one comparison operation based on the comparison between the decision tree and the expert assessment.