Machine learning-based diagnostic classifier

Abstract

Systems and methods for utilizing machine learning to generate a trans-diagnostic classifier that is operative to concurrently diagnose a plurality of different mental health disorders using a single trans-diagnostic questionnaire that includes a plurality of questions (e.g., 17 questions). Machine learning techniques are used to process labeled training data to build statistical models that include trans-diagnostic item-level questions as features to create a screen to classify groups of subjects as either healthy or as possibly having a mental health disorder. A subset of questions are selected from the multiple self-administered mental health questionnaires and used to autonomously screen subjects across multiple mental health disorders without physician involvement, optionally remotely and repeatedly, in a short amount of time.

What is claimed is: 
     
         1 . A system for screening the mental health of patients, the system comprising:
 a display;   a microphone;   a camera positioned to capture an image in front of the display;   a memory containing machine readable medium comprising machine executable code having stored thereon instructions for performing a method of evaluating the mental health of a user; and   a control system coupled to the memory comprising one or more processors, the control system configured to execute the machine executable code to cause the control system to:
 display, on the display, a set of text; 
 record, by the camera, a set of test video data which includes images of the user while reading the set of text aloud; 
 record, by the microphone, a set of test audio data which includes sounds of the user while reading the set of text aloud; 
 process the video data to output a set of video features comprising facial expressions of the user while reading the set of text aloud; and 
 process the audio data to output a set of audio features comprising tone of voice of the use while reading the set of text aloud; 
 output a mental health indication of the user by processing, using a machine learning model, the set of video features and the set of audio features, wherein the machine learning model was generated by:
 receiving labeled training data for a plurality of individuals, the labeled training data comprising audio and video data recorded for each of the plurality of individuals; 
 determining a plurality of features from the labeled training data; 
 training an initial machine learning model using a combination of the audio data and the video data of the labeled training data; 
 using the trained initial machine learning model to generate a plurality of subset machine learning models based on the plurality of features; and 
 selecting at least one of the subset machine learning models as the machine learning model. 
 
   
     
     
         2 . The system of  claim 1 , wherein the control system is further configured to execute the machine executable code to cause the control system to:
 execute a test application, by the control system, upon receiving, from the user interface, an indication to initiate a test; and   terminate the test application upon receiving, by the control system, an indication to stop the test,   wherein the indication to stop the test application comprises a determination, by the control system, that a user face is not within an image captured by the camera.   
     
     
         3 . The system of  claim 1 , wherein recording, by the microphone, further comprises:
 initiating the recording upon determining, by the control system, that the user is speaking.   
     
     
         4 . The system of  claim 1 , wherein the control system is further configured to execute the machine executable code to cause the control system to:
 receive the set of test video data and the set of test audio data;   preprocess the received set of test video data to identify a plurality of video segments, each video segment corresponding to one phrase in the set of text and comprising a time window; and   preprocess the received set of test audio data to identify a plurality of audio segments, each audio segment corresponding to one question in a series of questions and comprising a time window.   
     
     
         5 . The system of  claim 4 , wherein the control system is further configured to execute the machine executable code to cause the control system to:
 preprocess the plurality of audio segments and the plurality of video segments to identify overlapping time windows;   output a set of integrated audio and video segments based on the identified overlapping time windows.   
     
     
         6 . The system of  claim 1 , wherein the machine learning model is selected from the group consisting of: a generalized linear model, a regression model, a logistical regression model, and a supervised machine learning classification model. 
     
     
         7 . The system of  claim 1 , wherein the mental health indication identifies a likelihood of the user having one of a plurality of mental health disorders, the plurality of mental health disorders comprising: a neuropsychiatric disorder, schizophrenia, and a bipolar disorder. 
     
     
         8 . The system of  claim 1 , wherein the mental health indication identifies whether the user is a patient or a healthy control. 
     
     
         9 . The system of  claim 1 , wherein selecting at least one of the subset machine learning models as the machine learning model is based on classification performance, wherein each of the subset machine learning models includes a subset of the plurality of features. 
     
     
         10 . The system of  claim 9 , wherein each of the subset machine learning models includes features determined as having importance values above a predetermined threshold. 
     
     
         11 . The system of  claim 9 , wherein each of the subset machine learning models includes a different combination of the plurality of features. 
     
     
         12 . The system of  claim 1 , wherein the set of text includes a series of questions from mental health questionnaires, each question including a prompt and an answer for the prompt. 
     
     
         13 . The system of  claim 12 , wherein each of the plurality of features determined from the labeled training data correspond to a question in the series of questions. 
     
     
         14 . The system of  claim 13 , wherein the labeled training data includes audio and video data recorded for each of the plurality of individuals while reading aloud the prompt and answer to the prompt for each of the questions in the series of questions. 
     
     
         15 . A system for screening the mental health of patients, the system comprising:
 a display;   a microphone;   a camera configured to capture an image in front of the display;   a memory containing machine readable medium comprising machine executable code having stored thereon instructions for performing a method; and   a control system coupled to the memory comprising one or more processors, the control system configured to execute the machine executable code to cause the control system to perform a test by:
 displaying a set of text on the display; 
 recording, by the camera, a set of test video data which includes images of the user while reading the set of text aloud; 
 displaying, on the display, live video data recorded by the camera capturing the user; 
 iteratively processing the set of test video data during the test to:
 determine whether each of a plurality of pixels of a face of the user are within a frame on the display; and 
 stop the test in response to determining the face is outside the frame; 
 
 processing the set of test video data to identify video features; 
 storing the video features in the memory, wherein the video features include facial expressions of the user while reading the set of text aloud; and 
 processing, using a machine learning model, the set of video features to output a mental health indication of the user, wherein the machine learning model was generated by:
 receiving labeled training data for a plurality of individuals indicating whether each of the plurality of individuals has one or more mental health disorders, the labeled training data comprising video data recorded for each of the plurality of individuals while reading the set of text aloud; 
 determining a plurality of features from the labeled training data; 
 training an initial machine learning model using a combination of the video data of the labeled training data; 
 using the trained initial machine learning model to generate a plurality of subset machine learning models based on the plurality of features; and 
 selecting at least one of the subset machine learning models as the machine learning model. 
 
   
     
     
         16 . The system of  claim 15 , wherein iteratively processing the set of test video data further comprises:
 preprocessing a set of test audio data with the test video data to identify overlapping time windows; and   outputting a set of integrated audio and video segments based on the identified overlapping time windows.   
     
     
         17 . The system of  claim 16 , wherein the machine learning model is selected from the group consisting of: a generalized linear model, a regression model, a logistical regression model, and a supervised machine learning classification model. 
     
     
         18 . A machine learning training system, comprising:
 at least one non-transitory processor-readable storage medium that stores at least one of processor-executable instructions or data; and   at least one processor communicatively coupled to the at least one non-transitory processor-readable storage medium, in operation, the at least one processor configured to:   receive labeled training data for a plurality of individuals, the labeled training data further comprising:
 video data and audio data recorded while each of the plurality of individuals read predetermined text from a digital display, wherein the video data identifies portions of the video data comprising the face of the individual and the audio data identifies sounds representing the voice of the individual while reading the text from the digital display; 
   process the audio data, and the video data to output a plurality of features;   train an initial machine learning model using a combination of the audio data and the video data of the labeled training data;   use the trained initial machine learning model to generate a plurality of subset machine learning models based on the plurality of features;   select at least one of the plurality of subset machine learning models as a diagnostic classifier; and   store ones of the features corresponding to the diagnostic classifier in the at least one non-transitory processor-readable storage medium.   
     
     
         19 . The machine learning system of  claim 18 , wherein the diagnostic classifier is configured to output a mental health indication identifying an individual as healthy or as having one of: (i) a general mental health issue, and (ii) a specific mental health issue. 
     
     
         20 . The machine learning system of  claim 18 , wherein the diagnostic classifier is configured to output a mental health indication identifying an individual as having either a first specific mental health disorder or a second specific mental health disorder. 
     
     
         21 . The machine learning system of  claim 18 , wherein the diagnostic classifier is configured output a mental health indication identifying a risk of developing a mental health disorder for an individual. 
     
     
         22 . The machine learning system of  claim 18 , wherein the labeled training data further comprises:
 for each individual in the plurality of individuals, an indication of at least one of the following: whether the individual is healthy, whether the individual has a general mental health issue, whether the individual has one or more specific mental health disorders, whether the individual is at risk of developing a general mental health issue, or whether the individual is at risk of developing one or more specific mental health disorders.   
     
     
         23 . The machine learning system of  claim 18 , wherein training the initial machine learning model further comprises using k-fold cross validation with logistic regression. 
     
     
         24 . The machine learning system of  claim 18 , wherein the labeled training data further comprises at least one of functional measurement data or physiological measurement data. 
     
     
         25 . The machine learning system of  claim 18 , wherein the at least one processor is further configured to:
 use the features of the diagnostic classifier as a screening tool to assess at least one of intermediate or end-point outcomes in at least one clinical trial testing for treatment responses.