Machine learning techniques for enhanced redirection recommendation using real-time adjustment

1 . A computer-implemented method comprising:

receiving, by one or more processors, a service request data object for a user that is associated with (i) a probabilistic diagnosis data object and (ii) a diagnostic point-in-time feature that identifies at least a real-time location of the user;

determining, by the one or more processors and based at least in part on a search radius around the real-time location of the user, a first provider data object comprising (i) a first static operational feature associated with a first provider profile and (ii) a first dynamic operational feature associated with the first provider profile;

determining, by the one or more processors, using a first machine learning model, and based at least in part on the first static operational feature and the probabilistic diagnosis data object, a first generic redirection score for the first provider profile;

in response to determining that the first generic redirection score fails to satisfy a first threshold, expanding, by the one or more processors, the search radius around the user to identify a second provider data object comprising (i) a second static operational feature associated with a second provider profile and (ii) a second dynamic operational feature associated with the second provider profile;

determining, by the one or more processors, using the first machine learning model, and based at least in part on the second static operational feature and the probabilistic diagnosis data object, a second generic redirection score for the second provider profile;

determining, by the one or more processors and using a second machine learning model that is distinct from the first machine learning model, and based at least in part on the diagnostic point-in-time feature and the second dynamic operational feature, a real-time redirection score for the second provider profile;

determining, by the one or more processors and based at least in part on the second generic redirection score and the real-time redirection score, a predicted redirection score for the second provider profile; and

in response to determining that the predicted redirection score satisfies a second threshold, navigating, by the one or more processors and via an integrated mapping service, the user to a physical location associated with the second provider data object by:

rendering of a prompt within a first client device corresponding to the second provider data object,

responsive to receiving a first acceptance of the prompt from the first client device, establishing an electronic handshake between the first client device and a second client device associated with the service request data object, and

responsive to receiving a second acceptance via the electronic handshake, rendering a navigation instruction prompt within a third client device corresponding to the user, wherein the navigation instruction prompt comprises a link to the integrated mapping service.

2 . The computer-implemented method of claim 1 , wherein the first static operational feature is determined by:

retrieving asynchronous provider feature data associated with the first provider profile at an asynchronous provider data retrieval time; and

determining the first static operational feature based at least in part on the asynchronous provider feature data.

3 . The computer-implemented method of claim 1 , wherein the first static operational feature is determined by:

retrieving on-the-fly provider feature data at a request receipt time associated with the service request data object; and

determining the first static operational feature based at least in part on the on-the-fly provider feature data.

4 . The computer-implemented method of claim 1 , wherein determining the first generic redirection score comprises:

determining, based at least in part on an output of processing the first static operational feature using the first machine learning model, a multi-dimensional evaluation vector for the first provider profile that describes a plurality of dimensional evaluation scores that respectively correspond to a plurality of provider evaluation dimensions; and

determining, based at least in part on the plurality of dimensional evaluation scores, the first generic redirection score.

5 . The computer-implemented method of claim 4 , wherein:

the plurality of provider evaluation dimensions comprises a relative expertise dimension; and

a dimensional evaluation score of the relative expertise dimension describes an expertise measure of a geographically proximate provider profile associated with the first provider profile, wherein the geographically proximate provider profile is based at least in part on the search radius.

6 . The computer-implemented method of claim 1 , wherein:

the second machine learning model comprises a recurrent neural network,

the recurrent neural network is configured to sequentially process a sequence of model inputs via a sequence of timesteps until a terminal timestep that is associated with a redirection score expectation time, and

the real-time redirection score is determined based at least in part on a hidden state value for the terminal timestep.

7 . The computer-implemented method of claim 6 , wherein a model input for an nth timestep comprises an nth diagnostic point-in-time feature of one or more diagnostic point-in-time features and an nth dynamic operational feature of one or more dynamic operational features.

8 . A system comprising:

one or more processors; and

at least one memory storing processor-executable instructions that, when executed by any one or more of the one or more processors, causes the one or more processors to perform operations comprising:

receiving a service request data object for a user that is associated with (i) a probabilistic diagnosis data object and (ii) a diagnostic point-in-time feature that identifies at least a real-time location of the user;

determining, based at least in part on a search radius around the real-time location of the user, a first provider data object comprising (i) a first static operational feature associated with a first provider profile and (ii) a first dynamic operational feature associated with the first provider profile;

determining, using a first machine learning model, and based at least in part on the first static operational feature and the probabilistic diagnosis data object, a first generic redirection score for the first provider profile;

in response to determining that the first generic redirection score fails to satisfy a first threshold, expanding the search radius around the user to identify a second provider data object comprising (i) a second static operational feature associated with a second provider profile and (ii) a second dynamic operational feature associated with the second provider profile;

determining, using the first machine learning model, and based at least in part on the second static operational feature and the probabilistic diagnosis data object, a second generic redirection score for the second provider profile;

determining, using a second machine learning model that is distinct from the first machine learning model, and based at least in part on the diagnostic point-in-time feature and the second dynamic operational feature, a real-time redirection score for the second provider profile;

determining, based at least in part on the second generic redirection score and the real-time redirection score, a predicted redirection score for the second provider profile; and

in response to determining that the predicted redirection score satisfies a second threshold, navigating, via an integrated mapping service, the user to a physical location associated with the second provider data object by:

rendering of a prompt within a first client device corresponding to the second provider data object,

responsive to receiving a first acceptance of the prompt from the first client device, establishing an electronic handshake between the first client device and a second client device associated with the service request data object, and

responsive to receiving a second acceptance via the electronic handshake, rendering a navigation instruction prompt within a third client device corresponding to the user, wherein the navigation instruction prompt comprises a link to the integrated mapping service.

9 . The system of claim 8 , wherein the first static operational feature is determined by:

retrieving asynchronous provider feature data associated with the first provider profile at an asynchronous provider data retrieval time; and

determining the first static operational feature based at least in part on the asynchronous provider feature data.

10 . The system of claim 8 , wherein the first static operational feature is determined by:

retrieving on-the-fly provider feature data at a request receipt time associated with the service request data object; and

determining the first static operational feature based at least in part on the on-the-fly provider feature data.

11 . The system of claim 8 , wherein determining the first generic redirection score comprises:

determining, based at least in part on an output of processing the first static operational feature using the first machine learning model, a multi-dimensional evaluation vector for the first provider profile that describes a plurality of dimensional evaluation scores that respectively correspond to a plurality of provider evaluation dimensions; and

determining, based at least in part on the plurality of dimensional evaluation scores, the first generic redirection score.

12 . The system of claim 11 , wherein:

the plurality of provider evaluation dimensions comprises a relative expertise dimension; and

a dimensional evaluation score of the relative expertise dimension describes an expertise measure of a geographically proximate provider profile associated with the first provider profile, wherein the geographically proximate provider profile is based at least in part on the search radius.

13 . The system of claim 8 , wherein:

the second machine learning model comprises a recurrent neural network,

the recurrent neural network is configured to sequentially process a sequence of model inputs via a sequence of timesteps until a terminal timestep that is associated with a redirection score expectation time, and

the real-time redirection score is determined based at least in part on a hidden state value for the terminal timestep.

14 . The system of claim 13 , wherein a model input for an nth timestep comprises an nth diagnostic point-in-time feature of one or more diagnostic point-in-time features and an nth dynamic operational feature of one or more dynamic operational features.

15 . One or more non-transitory computer-readable storage media including instructions that, when executed by one or more processors, cause the one or more processors to perform operations, the operations comprising:

receiving a service request data object for a user that is associated with (i) a probabilistic diagnosis data object and (ii) a diagnostic point-in-time feature that identifies at least a real-time location of the user;

determining, based at least in part on a search radius around the real-time location of the user, a first provider data object comprising (i) a first static operational feature associated with a first provider profile and (ii) a first dynamic operational feature associated with the first provider profile;

determining, using a first machine learning model, and based at least in part on the first static operational feature and the probabilistic diagnosis data object, a first generic redirection score for the first provider profile;

in response to determining that the first generic redirection score fails to satisfy a first threshold, expanding the search radius around the user to identify a second provider data object comprising (i) a second static operational feature associated with a second provider profile and (ii) a second dynamic operational feature associated with the second provider profile;

determining, using the first machine learning model, and based at least in part on the second static operational feature and the probabilistic diagnosis data object, a second generic redirection score for the second provider profile;

determining, using a second machine learning model that is distinct from the first machine learning model, and based at least in part on the diagnostic point-in-time feature and the second dynamic operational feature, a real-time redirection score for the second provider profile;

determining, based at least in part on the second generic redirection score and the real-time redirection score, a predicted redirection score for the second provider profile; and

in response to determining that the predicted redirection score satisfies a second threshold, navigating, via an integrated mapping service, the user to a physical location associated with the second provider data object by:

rendering of a prompt within a first client device corresponding to the second provider data object,

responsive to receiving a first acceptance of the prompt from the first client device, establishing an electronic handshake between the first client device and a second client device associated with the service request data object, and

responsive to receiving a second acceptance via the electronic handshake, rendering a navigation instruction prompt within a third client device corresponding to the user, wherein the navigation instruction prompt comprises a link to the integrated mapping service.

16 . The one or more non-transitory computer-readable storage media of claim 15 , wherein the first static operational feature is determined by:

retrieving asynchronous provider feature data associated with the first provider profile at an asynchronous provider data retrieval time; and

determining the first static operational feature based at least in part on the asynchronous provider feature data.

17 . The one or more non-transitory computer-readable storage media of claim 15 , wherein the first static operational feature is determined by:

retrieving on-the-fly provider feature data at a request receipt time associated with the service request data object; and

determining the first static operational feature based at least in part on the on-the-fly provider feature data.

18 . The one or more non-transitory computer-readable storage media of claim 15 , wherein determining the first generic redirection score comprises:

determining, based at least in part on an output of processing the first static operational feature using the first machine learning model, a multi-dimensional evaluation vector for the first provider profile that describes a plurality of dimensional evaluation scores that respectively correspond to a plurality of provider evaluation dimensions; and

determining, based at least in part on the plurality of dimensional evaluation scores, the first generic redirection score.

19 . The one or more non-transitory computer-readable storage media of claim 18 , wherein:

the plurality of provider evaluation dimensions comprises a relative expertise dimension; and

a dimensional evaluation score of the relative expertise dimension describes an expertise measure of a geographically proximate provider profile associated with the first provider profile, wherein the geographically proximate provider profile is based at least in part on the search radius.

20 . The computer-implemented method of claim 1 , wherein the first machine learning model comprises a multi-output regression machine learning model and the second machine learning model comprises a recurrent neural network machine learning model.