Generating brain network impact scores

Abstract

Methods, systems, and apparatus, including computer programs encoded on computer storage media, for generating a mental health prediction for a patient. One of the methods includes obtaining brain data captured by one or more sensors characterizing a brain of a patient; determining a network graph from the brain data, wherein: a node of the network graph corresponds to a parcellation in the brain of the patient, and a subset of nodes of the network graph corresponds to a particular functional area of the brain; generating, for each of a plurality of nodes in the network graph, a measure of centrality of the node; generating, for the particular functional area of the brain and using the generated measures of centrality, a health score representing a measure of health of the functional area of the brain; generating a mental health prediction for the patient using the health score.

What is claimed is: 
     
         1 . A method comprising:
 obtaining brain data captured by one or more sensors characterizing a brain of a patient, wherein the brain data comprises, for each of a plurality of pairs of parcellations formed from a set of parcellations where each pair comprises a first parcellation and a second parcellation, data characterizing a number of tracts connecting the first parcellation and the second parcellation;   determining a network graph from the brain data, wherein:
 a node of the network graph corresponds to a parcellation in the brain of the patient, and 
 a subset of nodes of the network graph corresponds to a particular functional area of the brain; 
   generating, for each of a plurality of nodes in the network graph, a measure of centrality of the node;   generating, for the particular functional area of the brain and using the generated measures of centrality, a health score representing a measure of health of the functional area of the brain; and   generating a mental health prediction for the patient using the health score.   
     
     
         2 . The method of  claim 1 , wherein:
 an edge between a first node and a second node of the network graph characterizes a connection between the parcellation corresponding to the first node and the parcellation corresponding to the second node, and   determining a network graph from the brain data comprises, for an edge between a first node and a second node of the network graph, generating an edge weight according to the number of tracts connecting the parcellation corresponding to the first node and the parcellation corresponding to the second node;   
     
     
         3 . The method of  claim 1 , wherein the edge weight for the edge between a first node and a second node in the network graph is generated according to i) the number of tracts connecting the parcellation corresponding to the first node and the parcellation corresponding to the second node and ii) a measure of health of the tracts connecting the parcellation corresponding to the first node and the parcellation corresponding to the second node. 
     
     
         4 . The method of  claim 3 , wherein the edge weight for the edge between the first node and the second node is a machine-learned linear combination of i) the number of tracts connecting the parcellation corresponding to the first node and the parcellation corresponding to the second node and ii) a measure of health of the tracts connecting the parcellation corresponding to the first node and the parcellation corresponding to the second node. 
     
     
         5 . The method of  claim 1 , further comprising generating, for each of a plurality of subsets of nodes of the network graph wherein each subset corresponds to a different functional area of the brain and using the generated measures of centrality, a respective health score representing a measure of health of the respective functional areas of the brain. 
     
     
         6 . The method of  claim 5 , wherein generating a mental health prediction for the patient comprises determining a relationship between the measures of health of a plurality of different functional areas of the brain. 
     
     
         7 . The method of  claim 1 , wherein generating a health score for a subset of parcellations comprises generating the health score using i) the generated measures of centrality, ii) the number of tracts connecting each pair of parcellations in the subset of parcellations, and iii) a measure of health of the tracts connecting each pair of parcellations in the subset of parcellations. 
     
     
         8 . The method of  claim 7 , wherein the health score for a subset of parcellations is a machine-learned linear combination of i) the generated measures of centrality, ii) the number of tracts connecting each pair of parcellations in the subset of parcellations, and iii) a measure of health of the tracts connecting each pair of parcellations in the subset of parcellations. 
     
     
         9 . The method of  claim 1 , wherein generating a mental health prediction for the patient comprises determining whether to perform a particular treatment on the brain of the patient. 
     
     
         10 . The method of  claim 9 , wherein the particular treatment is associated with a range of possible health of scores of the subset of parcellations. 
     
     
         11 . The method of  claim 10 , wherein the range of possible health scores associated with the particular treatment is machine learned, the learning comprising processing a plurality of training examples corresponding to respective other patients. 
     
     
         12 . The method of  claim 1 , wherein generating a mental health prediction for the patient comprises generating a prediction of whether the patient has a particular brain disease. 
     
     
         13 . The method of  claim 12 , wherein the particular brain disease is associated with a range of possible health of scores of the subset of parcellations. 
     
     
         14 . The method of  claim 13 , wherein the range of possible health scores associated with the particular brain disease is machine learned, the learning comprising processing a plurality of training examples corresponding to respective other patients. 
     
     
         15 . The method of  claim 1 , wherein:
 the obtained brain data corresponds to a first time point;   the health score is a first health score that represents a measure of health of the functional area of the brain at the first time point;   the method further comprises:
 obtaining second brain data corresponding to a second time point, and 
 generating, for the particular functional area of the brain, a second health score representing a measure of health of the functional area of the brain at the second time point; and 
   generating a mental health prediction for the patient comprises determining a change between the first health score and the second health score.   
     
     
         16 . The method of  claim 15 , wherein:
 the first time point is a time point before a treatment, and   the second time point is a time point after the treatment.   
     
     
         17 . The method of  claim 16 , wherein the treatment comprises one or more of:
 a drug therapy;   radiation therapy;   magnetic therapy;   light therapy;   ultrasonic therapy;   gene-modifying therapy;   a surgery;   physical therapy; or   mental exercises.   
     
     
         18 . The method of  claim 16 , wherein the mental health prediction comprises one or more of:
 a measure of efficacy of the treatment; or   a performance measure of one or more entities that administered the treatment.   
     
     
         19 . The method of  claim 1 , wherein generating a mental health prediction comprises generating a sequence of treatments, wherein each treatment in the sequence of treatments corresponds to a respective range of health scores. 
     
     
         20 . A system comprising one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or more computers, to cause the one or more computers to perform operations comprising:
 obtaining brain data captured by one or more sensors characterizing a brain of a patient, wherein the brain data comprises, for each of a plurality of pairs of parcellations formed from a set of parcellations where each pair comprises a first parcellation and a second parcellation, data characterizing a number of tracts connecting the first parcellation and the second parcellation;   determining a network graph from the brain data, wherein:
 a node of the network graph corresponds to a parcellation in the brain of the patient, and 
 a subset of nodes of the network graph corresponds to a particular functional area of the brain; 
   generating, for each of a plurality of nodes in the network graph, a measure of centrality of the node;   generating, for the particular functional area of the brain and using the generated measures of centrality, a health score representing a measure of health of the functional area of the brain; and   generating a mental health prediction for the patient using the health score.