Gratitude prediction machine learning models
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for determining gratitude scores for a plurality of people who have interacted with an organization. In one aspect, a method comprises: obtaining, for each of a plurality of people who have interacted with an organization, history data characterizing previous interactions of the person with the organization and third party data characterizing aspects of the person outside of their previous interactions with the organization; and processing, for each of the plurality of people, the history data and the third party data for the person using a gratitude prediction machine learning model to generate a gratitude score for the person, wherein the gratitude score for the person characterizes a likelihood that the person will take a specified action on behalf of the organization in the future.
1. A method performed by one or more data processing apparatus, the method comprising:
obtaining a plurality of training examples, wherein each training example corresponds to a respective person that has interacted with an organization, and each training example comprises: (i) a training input comprising history data characterizing previous interactions of the corresponding person with the organization, and (ii) a target gratitude score that characterizes whether the corresponding person has taken a specified action on behalf of the organization;
training a gratitude prediction machine learning model having a plurality of model parameters on the plurality of training examples, wherein:
the gratitude prediction machine learning model is configured to process history data for an input person in accordance with values of the plurality of model parameters of the gratitude prediction machine learning model to generate a gratitude score for the input person;
the gratitude score for the input person characterizes a likelihood that the input person will take the specified action on behalf of the organization in the future;
training the gratitude prediction machine learning model comprises determining trained values of the model parameters of the gratitude prediction machine learning model from initial values of the model parameters of the gratitude prediction machine learning model by a machine learning training technique; and
training the gratitude prediction machine learning model on the plurality of training examples comprises, for each training example, training the gratitude prediction machine learning model to process the training input of the training example to generate an output gratitude score that matches the target gratitude score for the training example;
after training the gratitude prediction machine learning model:
generating a respective gratitude score for each of a plurality of people using the trained machine learning model;
determining a ranking of the plurality of people based at least in part on their respective gratitude scores; and
taking an action based on the ranking of the plurality of people, comprising:
processing the ranking of the plurality of people based on their respective gratitude scores to automatically generate a communication list that designates a subset of the plurality of people; and
automatically transmitting communications using the subset of the plurality of people that are designated by the communication list;
determining that a criterion for re-training the gratitude prediction machine learning model has been satisfied, comprising determining that the gratitude prediction machine learning model has not been re-trained for at least a threshold duration of time; and
in response:
determining an updated set of training examples for training the gratitude prediction machine learning model wherein the updated set of training examples include at least one new training example that was not previously used for training the gratitude prediction machine learning model;
re-training the gratitude prediction machine learning model on the updated set of training examples; and
determining an updated ranking of the plurality of people using the re-trained gratitude prediction machine learning model.
2. The method of claim 1 , wherein the gratitude prediction machine learning model is configured to process both history data and third party data for the input person, wherein the third party data characterizes aspects of the input person outside of their previous interactions with the organization.
3. The method of claim 1 , wherein the third party data characterizes actions taken by the input person on behalf of other organizations, actions related to the organization that are taken by the input person on a social network, or both.
4. The method of claim 1 , wherein the ranking of the plurality of people identifies one or more people with the highest gratitude scores from among the plurality of people.
5. The method of claim 1 , wherein the ranking of the plurality of people identifies each person having a gratitude score that exceeds a predefined threshold.
6. The method of claim 1 , wherein the ranking of the plurality of people defines an ordering of the plurality of people based on their respective gratitude scores.
7. The method of claim 6 , wherein the ordering is a highest-to-lowest ordering.
8. The method of claim 1 , wherein the ranking of the plurality of people classifies each person into a set of categories including a high gratitude category and a low gratitude category.
9. The method of claim 1 , further comprising
determining, for each of the plurality of people, a respective measure of resources associated with the person;
wherein the ranking of the plurality of people is further based at least in part on the respective measure of resources associated with each person.
10. The method of claim 9 , further comprising:
determining, for each of the plurality of people, a respective composite score for the person based on: (i) the gratitude score for the person, and (ii) the measure of resources associated with the person;
wherein the ranking of the plurality of people is based on the respective composite score for each person.
11. The method of claim 9 , wherein the ranking of the plurality of people classifies each person into a set of categories including a high resource—high gratitude category, a low resource—low gratitude category, a low resource—high gratitude category, and a high resource—low gratitude category.
12. The method of claim 1 , wherein the organization is a healthcare organization, and the gratitude prediction machine learning model is configured to process history data for the input person that characterizes previous interactions of the input person with the organization in relation to healthcare services provided to the input person by the organization.
13. The method of claim 1 , wherein the history data for the input person characterizes one or more of: appointments where the input person received healthcare services from the organization, compensation received by the organization for providing healthcare services to the input person, hospitals affiliated with the organization where the input person has received healthcare services, hospital departments where the input person has received healthcare services, or healthcare providers who have contributed to providing healthcare services to the input person.
14. The method of claim 1 , wherein the organization is a healthcare organization, and the gratitude prediction machine learning model is configured to process history data for the input person that characterizes previous interactions of the input person with the organization outside of the provision of healthcare services to the input person by the organization.
15. The method of claim 14 , wherein the history data for the input person characterizes one or more of: responses provided by the input person to surveys issued by the organization, or
previous actions taken by the input person on behalf of the organization.
16. A system comprising:
one or more computers; and
one or more storage devices communicatively coupled to the one or more computers, wherein the one or more storage devices store instructions that, when executed by the one or more computers, cause the one or more computers to perform operations comprising:
obtaining a plurality of training examples, wherein each training example corresponds to a respective person that has interacted with an organization, and each training example comprises: (i) a training input comprising history data characterizing previous interactions of the corresponding person with the organization, and (ii) a target gratitude score that characterizes whether the corresponding person has taken a specified action on behalf of the organization;
training a gratitude prediction machine learning model having a plurality of model parameters on the plurality of training examples, wherein:
the gratitude prediction machine learning model is configured to process history data for an input person in accordance with values of the plurality of model parameters of the gratitude prediction machine learning model to generate a gratitude score for the input person;
the gratitude score for the input person characterizes a likelihood that the input person will take the specified action on behalf of the organization in the future;
training the gratitude prediction machine learning model comprises determining trained values of the model parameters of the gratitude prediction machine learning model from initial values of the model parameters of the gratitude prediction machine learning model by a machine learning training technique; and
training the gratitude prediction machine learning model on the plurality of training examples comprises, for each training example, training the gratitude prediction machine learning model to process the training input of the training example to generate an output gratitude score that matches the target gratitude score for the training example;
after training the gratitude prediction machine learning model:
generating a respective gratitude score for each of a plurality of people using the trained machine learning model;
determining a ranking of the plurality of people based at least in part on their respective gratitude scores; and
taking an action based on the ranking of the plurality of people, comprising:
processing the ranking of the plurality of people based on their respective gratitude scores to automatically generate a communication list that designates a subset of the plurality of people; and
automatically transmitting communications using the subset of the plurality of people that are designated by the communication list;
determining that a criterion for re-training the gratitude prediction machine learning model has been satisfied, comprising determining that the gratitude prediction machine learning model has not been re-trained for at least a threshold duration of time; and
in response:
determining an updated set of training examples for training the gratitude prediction machine learning model, wherein the updated set of training examples include at least one new training example that was not previously used for training the gratitude prediction machine learning model;
re-training the gratitude prediction machine learning model on the updated set of training examples; and
determining an updated ranking of the plurality of people using the re-trained gratitude prediction machine learning model.
17. One or more non-transitory computer storage media storing instructions that when executed by one or more computers cause the one or more computers to perform operations comprising:
obtaining a plurality of training examples, wherein each training example corresponds to a respective person that has interacted with an organization, and each training example comprises: (i) a training input comprising history data characterizing previous interactions of the corresponding person with the organization, and (ii) a target gratitude score that characterizes whether the corresponding person has taken a specified action on behalf of the organization;
training a gratitude prediction machine learning model having a plurality of model parameters on the plurality of training examples, wherein:
the gratitude prediction machine learning model is configured to process history data for an input person in accordance with values of the plurality of model parameters of the gratitude prediction machine learning model to generate a gratitude score for the input person;
the gratitude score for the input person characterizes a likelihood that the input person will take the specified action on behalf of the organization in the future;
training the gratitude prediction machine learning model comprises determining trained values of the model parameters of the gratitude prediction machine learning model from initial values of the model parameters of the gratitude prediction machine learning model by a machine learning training technique; and
training the gratitude prediction machine learning model on the plurality of training examples comprises, for each training example, training the gratitude prediction machine learning model to process the training input of the training example to generate an output gratitude score that matches the target gratitude score for the training example;
after training the gratitude prediction machine learning model:
generating a respective gratitude score for each of a plurality of people using the trained machine learning model;
determining a ranking of the plurality of people based at least in part on their respective gratitude scores; and
taking an action based on the ranking of the plurality of people, comprising:
processing the ranking of the plurality of people based on their respective gratitude scores to automatically generate a communication list that designates a subset of the plurality of people; and
automatically transmitting communications using the subset of the plurality of people that are designated by the communication list;
determining that a criterion for re-training the gratitude prediction machine learning model has been satisfied, comprising determining that the gratitude prediction machine learning model has not been re-trained for at least a threshold duration of time; and
in response:
determining an updated set of training examples for training the gratitude prediction machine learning model, wherein the updated set of training examples include at least one new training example that was not previously used for training the gratitude prediction machine learning model;
re-training the gratitude prediction machine learning model on the updated set of training examples; and
determining an updated ranking of the plurality of people using the re-trained gratitude prediction machine learning model.