Predicting likelihoods of conditions being satisfied using neural networks
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for predicting likelihoods of conditions being satisfied using recurrent neural networks. One of the systems is configured to process a temporal sequence comprising a respective input at each of a plurality of time steps and comprises: one or more recurrent neural network layers; one or more logistic regression nodes, wherein each of the logistic regression nodes corresponds to a respective condition from a predetermined set of conditions, and wherein each of the logistic regression nodes is configured to, for each of the plurality of time steps: receive the network internal state for the time step; and process the network internal state for the time step in accordance with current values of a set of parameters of the logistic regression node to generate a future condition score for the corresponding condition for the time step.
1. A method comprising:
obtaining a first temporal sequence of health events, wherein the first temporal sequence comprises respective health-related data associated with a particular patient at each of a plurality of time steps,
for each of the plurality of time steps, processing the respective health-related data associated with the particular patient at the time step using a neural network, wherein the processing comprises updating a respective internal state of each neural network layer of the neural network using the respective health-related data associated with the particular patient at the time step to generate a network internal state of the neural network for the time step;
generating, from the network internal state of the neural network after a last time step in the first temporal sequence, a neural network output for the first temporal sequence, wherein the network internal state after the last time step is an updated internal state for the last neural network layer in the neural network for the last health event in the first temporal sequence, and
generating, from the neural network output for the first temporal sequence, health analysis data that characterizes future health events that may occur after the last time step in the first temporal sequence.
2. The method of claim 1 , wherein, for one or more of the time steps, the health-related data at the time step is a respective token from a predetermined vocabulary of tokens, each token in the predetermined vocabulary of tokens representing a different health event.
3. The method of claim 2 , wherein, for one or more of the time steps, the health-related data at the time step is other health-related data classified as impacting health of the particular patient.
4. The method of claim 2 , wherein obtaining the first temporal sequence comprises:
accessing an electronic medical record for the particular patient;
identifying health events in the electronic medical record;
determining, for each health event identified in the electronic medical record, a token in the predetermined vocabulary of tokens that represents the health event; and
generating a temporal sequence that includes the tokens that represent the identified health events ordered by time that corresponding health events occurred.
5. The method of claim 1 , wherein the neural network further comprises an output layer that is trained to process the network internal state for the last time step to generate the neural network output, wherein the neural network output comprises a respective score for each of a plurality of possible health events, wherein the respective score for each of the possible health events represents a likelihood that the possible health event is a health event at a time step subsequent to the last time step in the first temporal sequence.
6. The method of claim 5 , wherein generating the health analysis data comprises generating data identifying one or more highest-scoring health events using the respective scores.
7. The method of claim 1 , further comprising:
obtaining data identifying an additional health event;
processing the additional health event using the neural network to generate a modified network internal state;
generating an updated network output from the modified network internal state; and
generating updated health analysis data from the updated network output.
8. A system comprising one or more computers and one or more storage devices storing instructions that when executed by the one or more computers cause the one or more computers to perform operations comprising:
obtaining a first temporal sequence of health events, wherein the first temporal sequence comprises respective health-related data associated with a particular patient at each of a plurality of time steps,
for each of the plurality of time steps, processing the respective health-related data associated with the particular patient at the time step using a neural network, wherein the processing comprises updating a respective internal state of each neural network layer of the neural network using the respective health-related data associated with the particular patient at the time step to generate a network internal state of the neural network for the time step;
generating, from the network internal state of the neural network after a last time step in the first temporal sequence, a neural network output for the first temporal sequence, wherein the network internal state after the last time step is an updated internal state for the last neural network layer in the neural network for the last health event in the first temporal sequence, and
generating, from the neural network output for the first temporal sequence, health analysis data that characterizes future health events that may occur after the last time step in the first temporal sequence.
9. The system of claim 8 , wherein, for one or more of the time steps, the health-related data at the time step is a respective token from a predetermined vocabulary of tokens, each token in the predetermined vocabulary of tokens representing a different health event.
10. The system of claim 9 , wherein, for one or more of the time steps, the health-related data at the time step is other health-related data classified as impacting health of the particular patient.
11. The system of claim 9 , wherein obtaining the first temporal sequence comprises:
accessing an electronic medical record for the particular patient;
identifying health events in the electronic medical record;
determining, for each health event identified in the electronic medical record, a token in the predetermined vocabulary of tokens that represents the health event; and
generating a temporal sequence that includes the tokens that represent the identified health events ordered by time that corresponding health events occurred.
12. The system of claim 8 , wherein the neural network further comprises an output layer that is trained to process the network internal state for the last time step to generate the neural network output, wherein the neural network output comprises a respective score for each of a plurality of possible health events, wherein the respective score for each of the possible health events represents a likelihood that the possible health event is a health event at a time step subsequent to the last time step in the first temporal sequence.
13. The system of claim 12 , wherein generating the health analysis data comprises generating data identifying one or more highest-scoring health events using the respective scores.
14. The system of claim 8 , wherein the operations further comprise:
obtaining data identifying an additional health event;
processing the additional health event using the neural network to generate a modified network internal state;
generating an updated network output from the modified network internal state; and
generating updated health analysis data from the updated network output.
15. A computer program product encoded on one or more non-transitory computer readable media, the computer program product comprising instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:
obtaining a first temporal sequence of health events, wherein the first temporal sequence comprises respective health-related data associated with a particular patient at each of a plurality of time steps,
for each of the plurality of time steps, processing the respective health-related data associated with the particular patient at the time step using a neural network, wherein the processing comprises updating a respective internal state of each neural network layer of the neural network using the respective health-related data associated with the particular patient at the time step to generate a network internal state of the neural network for the time step;
generating, from the network internal state of the neural network after a last time step in the first temporal sequence, a neural network output for the first temporal sequence, wherein the network internal state after the last time step is an updated internal state for the last neural network layer in the neural network for the last health event in the first temporal sequence, and
generating, from the neural network output for the first temporal sequence, health analysis data that characterizes future health events that may occur after the last time step in the first temporal sequence.
16. The computer program product of claim 15 , wherein, for one or more of the time steps, the health-related data at the time step is a respective token from a predetermined vocabulary of tokens, each token in the predetermined vocabulary of tokens representing a different health event.
17. The computer program product of claim 16 , wherein, for one or more of the time steps, the health-related data at the time step is other health-related data classified as impacting health of the particular patient.
18. The computer program product of claim 16 , wherein obtaining the first temporal sequence comprises:
accessing an electronic medical record for the particular patient;
identifying health events in the electronic medical record;
determining, for each health event identified in the electronic medical record, a token in the predetermined vocabulary of tokens that represents the health event; and
generating a temporal sequence that includes the tokens that represent the identified health events ordered by time that corresponding health events occurred.
19. The computer program product of claim 15 , wherein the neural network further comprises an output layer that is trained to process the network internal state for the last time step to generate the neural network output, wherein the neural network output comprises a respective score for each of a plurality of possible health events, wherein the respective score for each of the possible health events represents a likelihood that the possible health event is a health event at a time step subsequent to the last time step in the first temporal sequence.
20. The computer program product of claim 15 , wherein the operations further comprise:
obtaining data identifying an additional health event;
processing the additional health event using the neural network to generate a modified network internal state;
generating an updated network output from the modified network internal state; and
generating updated health analysis data from the updated network output.