Subrogation case management

1. A computer-implemented method for facilitating automated subrogation case processing, comprising:

transmitting a query to a database using a communication subsystem, the query including information indicative of an incident associated with a member;

receiving, by the communication subsystem, a response to the query indicative of associated medical information for the member;

retrieving, based on the received response, medical claims associated with the member;

computing, using a computation subsystem that is at least partially implemented using electronic circuits, a first probability value for each retrieved medical claim, the first probability indicating a relatedness of the medical claim to the incident;

determining, for each retrieved medical claim, whether the medical claim is related to the incident by comparing the first probability value with a first predefined threshold;

computing, using the computation subsystem, a second probability value indicating whether all medical claims related to the incident are present; and

generating a final determination for the incident by comparing the second probability value with a second predefined threshold, wherein the method further comprises:

obtaining a first set of training data that includes information of multiple past incidents, each of the multiple past incidents having one or more related medical claims and each medical claim including at least an age of a corresponding member, one or more diagnosis codes, one or more drug codes, or one or more procedural codes; and

training, by the computation subsystem, a claim valuation model based on at least one of the age of the corresponding member, the one or more diagnosis codes, the one or more drug codes, or the one or more procedural codes of the first set of training data,

wherein the training comprises training the claim valuation model using a gradient boosting technique.

2. The method of claim 1 , wherein computing the first probability value comprises:

determining the first probability value by obtaining an inference result of the medical claim using the claim valuation model.

3. The method of claim 1 , wherein the gradient boosting technique uses a decision tree having a maximum depth of five.

4. The method of claim 1 , comprising:

obtaining a second set of training data that includes information of multiple past incidents, each of the multiple past incidents including at least an age of a corresponding member, a sum of paid amount for all related medical claims, a number of unrelated medical claims, or a sum of paid amount for all the unrelated medical claims; and

training, by the computation subsystem, a final recommendation model based on at least one of the age of the corresponding member, the sum of paid amount for all the related medical claims, the number of the unrelated medical claims, or the sum of paid amount for all the unrelated medical claims in the second set of training data.

5. The method of claim 4 , wherein computing the second probability value comprises:

determining the second probability value by obtaining an inference result based on the incident and the retrieved medical claims using the final recommendation model.

6. The method of claim 4 , wherein the training comprises training the final recommendation model using a gradient boosting technique.

7. The method of claim 6 , wherein the gradient boosting techniques uses a decision tree having a maximum depth of five.

8. A device for automated subrogation case processing, the device comprising:

a processor; and

a memory including instruction stored thereon, wherein the instructions upon execution by the processor cause the processor to:

transmit a query to a database, the query including information indicative of an incident associated with a member;

receive a response to the query indicative of an associated medical information for the member;

retrieve, based on the received response, medical claims associated with the member;

compute, for each retrieved medical claim, a first probability value indicating a relatedness of the medical claim to the incident;

determine, for each retrieved medical claim, whether the medical claim is related to the incident by comparing the first probability value with a first predefined threshold;

compute a second probability value indicating whether all medical claims related to the incident are present; and

generate a final determination for the incident by comparing the second probability value with a second predefined threshold,

wherein the processor is further configured to:

receive a set of training data that includes information of multiple past incidents, each of the multiple past incidents including at least an age of a corresponding member, a sum of paid amount for all related medical claims, a number of unrelated medical claims, or a sum of paid amount for all the unrelated medical claims;

train a final recommendation model based on at least one of the age of the corresponding member, the sum of paid amount for all the related medical claims, the number of the unrelated medical claims, or the sum of paid amount for all the unrelated medical claims in the set of training data, and

the processor is configured to compute the second probability value based on:

determining an inference result based on the incident and the retrieved medical claims using the final recommendation model.

9. The device of claim 8 , wherein the processor is configured to:

receive an additional set of training data that includes information of multiple past incidents, each of the multiple past incidents having one or more related medical claims and each medical claim including at least an age of a corresponding member, one or more diagnosis codes, one or more drug codes, or one or more procedural codes; and

train a claim valuation model based on at least one of the age of the corresponding member, the one or more diagnosis codes, the one or more drug codes, or the one or more procedural codes of the additional set of training data.

10. The device of claim 9 , wherein the processor is configured to compute the first probability value based on:

determining an inference result of the medical claim using the claim valuation model.

11. The device of claim 10 , wherein the processor is configured to train the claim valuation model using a gradient boosting technique with a decision tree that has a maximum depth of five.

12. The device of claim 8 , wherein the processor is configured to compute the first probability value based on:

determining an inference result based on the incident and the retrieved medical claims using the final recommendation model.

13. The device of claim 8 , wherein the processor is configured to train the claim valuation model using a gradient boosting technique with a decision tree that has a maximum depth of five.

14. A system for automated subrogation case processing, comprising

one or more processors; and

one or more memory units including instruction stored thereon, wherein the instructions upon execution by the one or more processors cause the one or more processors to implement a subrogation case management platform including:

a presentation-layer module configured to provide a user-interface for a member of the platform to interact and retrieve information;

a service-layer module in communication with the presentation-layer module to obtain information indicative of an incident associated with the member via the user-interface, wherein the service-layer module comprises an address verification module, a member authorization module, a notification module, and a data layer module, the service-layer module further comprising:

a case-management service configured to communicate with the address verification module to verify an address of the member, to communicate with the medical authorization module to obtain a medical authorization from the member, to communicate with the notification module to notify the member of a change in processing the incident, and to communicate with the data layer to retrieve medical claims of the member from one or more databases based on the information indicative of the incident, the case-management service further comprising a computation subsystem configured to:

compute, for each retrieved medical claim, a first probability value indicating a relatedness of the medical claim to the incident;

determine, for each retrieved medical claim, whether the medical claim is related to the incident by comparing the first probability value with a first predefined threshold;

compute a second probability value indicating whether all medical claims related to the incident are present; and

generate a final determination for the incident by comparing the second probability value with a second predefined threshold,

wherein the case-management service is further configured to:

train a claim valuation model based on a first set of input data;

train a final recommendation model based on a second set of input data; and

deploy the claim valuation model and the final recommendation model upon determining that the claim valuation model and the final recommendation model achieve a predefined level of accuracy; and wherein the case-management service further comprises a job scheduler configured to retrain the claim valuation model or the final recommendation model.