Generating templated documents using machine learning techniques

Patent metadata

Field	Value
Publication number	US-10691998-B2
Application number	US-201615385804-A
Country	US
Kind code	B2
Filing date	Dec 20, 2016
Priority date	Dec 20, 2016
Publication date	Jun 23, 2020
Grant date	Jun 23, 2020

Abstract

Systems and methods of predicting documentation associated with an encounter between attendees are provided. For instance, attendee data indicative of one or more previous visit notes associated with a first attendee can be obtained. The attendee data can be inputted into a machine-learned note prediction model that includes a neural network. The neural network can generate one or more context vectors descriptive of the attendee data. Data indicative of a predicted visit note can be received as output of the machine-learned note prediction model based at least in part on the context vectors. The predicted visit note can include a set of predicted information expected to be included in a subsequently generated visit note associated with the first attendee.

First claim

What is claimed is:

1. A computer-implemented method of predicting documentation associated with an encounter between attendees, the method comprising:
obtaining, by one or more computing devices, attendee data indicative of one or more previously generated visit notes associated with a first attendee of a subject encounter between the first attendee and a second attendee;
inputting, by the one or more computing devices, the attendee data into a machine-learned note prediction model comprising a first neural network and a second neural network;
receiving, by the one or more computing devices, one or more context vectors as output of the first neural network;
inputting, by the one or more computing devices, the one or more context vectors into the second neural network of the machine-learned note prediction model;
receiving, by the one or more computing devices, one or more prediction vectors as output of the second neural network, the one or more prediction vectors being descriptive of information to potentially be included in a predicted visit note;
providing, by the one or more computing devices, the one or more prediction vectors as input to a suggestion model of the note prediction model;
providing, by the one or more computing devices, data indicative of a first text entry input by a user as input to the suggestion model; and
receiving as output of the machine-learned note prediction model, by the one or more computing devices, data indicative of a predicted visit note, the predicted visit note comprising a set of predicted information expected to be included in a subsequently generated visit note associated with the first attendee, the set of predicted information comprising one or more suggested text entries determined based at least in part on the one or more prediction vectors and the data indicative of the first text entry.

2. The computer-implemented method of claim 1 , wherein the attendee data comprises data indicative of one or more previously generated visit notes for the first attendee, each previously generated visit note being associated with a previous encounter of the first attendee.

3. The computer-implemented method of claim 1 , wherein the attendee data further comprises data associated with the subject encounter between the first attendee and the second attendee.

4. The computer-implemented method of claim 3 , wherein the data associated with the subject encounter comprises data provided to a user computing device prior to a generation of a visit note associated with the subject encounter.

5. The computer-implemented method of claim 1 , wherein the first attendee is a patient associated with the subject encounter and the second attendee is a doctor associated the subject encounter, and wherein the attendee data includes data relating to a medical history of the patient.

6. The computer-implemented method of claim 5 , wherein the predicted information comprises substantive information expected to be included in a subsequently generated visit note associated with the subject encounter based at least in part on the attendee data.

7. The computer-implemented method of claim 1 , wherein the first neural network comprises a long short-term memory recurrent neural network.

8. The computer-implemented method of claim 1 , wherein receiving, by the one or more computing devices, data indicative of a predicted visit note comprises receiving the data indicative of the predicted visit note as an output of the second neural network.

9. The computer-implemented method of claim 1 , further comprising training, by the one or more computing devices, the note prediction model based on a set of training data;
wherein training, by the one or more computing devices, the note prediction model comprises backpropagating, by the one or more computing devices, a loss function through the note prediction model.

10. The computer-implemented method of claim 9 , wherein the training data comprises data indicative of a plurality of global visit notes.

11. The computer-implemented method of claim 9 , wherein the training data comprises data indicative of a plurality of doctor specific visit notes.

12. A computing system, comprising:
one or more processors; and
one or more memory devices, the one or more memory devices storing computer-readable instructions that when executed by the one or more processors cause the one or more processors to perform operations, the operations comprising:
obtaining attendee data indicative of one or more previous visit notes associated with a first attendee of a subject encounter between the first attendee and a second attendee;
inputting the attendee data into a machine-learned note prediction model comprising a first neural network and a second neural network;

receiving, by the one or more computing devices, one or more context vectors as output of the first neural network;
inputting, by the one or more computing devices, the one or more context vectors into the second neural network of the machine-learned note prediction model;
receiving, by the one or more computing devices, one or more prediction vectors as output of the second neural network, the one or more prediction vectors being descriptive of information to potentially be included in a predicted visit note;
providing, by the one or more computing devices, the one or more prediction vectors as input to a suggestion model of the note prediction model;
providing, by the one or more computing devices, data indicative of a first text entry input by a user as input to the suggestion model; and
receiving, as output of the machine-learned note prediction model, data indicative of a predicted visit note, the predicted visit note comprising a set of predicted information expected to be included in a subsequently generated visit note associated with the subject encounter, the set of predicted information comprising one or more suggested text entries determined based at least in part on the one or more prediction vectors and the data indicative of the first text entry.

13. The computing system of claim 12 , wherein the attendee data comprises data indicative of one or more previously generated visit notes for the first attendee, each previously generated visit note being associated with a previous encounter of the first attendee.

14. One or more tangible, non-transitory computer-readable media storing computer-readable instructions that when executed by one or more processors cause the one or more processors to perform operations, the operations comprising
obtaining patient data indicative of one or more previous visit notes associated with a patient;
inputting the patient data into a first neural network associated with a machine-learned note prediction model;
receiving one or more context vectors as output of the first neural network;
inputting the one or more context vectors into a second neural network associated with the machine-learned note prediction model;
receiving one or more prediction vectors as output of the second neural network, the one or more prediction vectors being descriptive of information to potentially be included in a predicted visit note;
providing the one or more prediction vectors as input to a suggestion model associated with the note prediction model;
providing data indicative of a first text entry input by a user as input to the suggestion model; and
receiving data indicative of a predicted visit note, the predicted visit note comprising a set of predicted information expected to be included in a subsequently generated visit note associated with the patient, the set of predicted information comprising one or more

Assignees

• Google Inc
• Google Llc

Inventors

• Po Ming Jack
• Co Christopher
• Chou Katherine

Classifications

• G16H15/00 Primary

  ICT specially adapted for medical reports, e.g. generation or transmission thereof · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06N3/0455

  Auto-encoder networks; Encoder-decoder networks · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06Q50/22 Primary

  Social work or social welfare, e.g. community support activities or counselling services · CPC title