Predictive and Prescriptive Analytics for Managing High-Cost Claimants in Healthcare
A mechanism is provided in a data processing system for predictive and prescriptive analytics for managing high-cost claimants. The mechanism trains a machine learning model using the set of de-identified claims data to predict high-cost claimants using the training data. The mechanism applies transfer learning by retraining the machine learning model using a first set of customized client data to generate a client-specific machine learning model. The mechanism then applies the client-specific machine learning model to a second set of customized client data to identify a set of predicted high-cost claimants within the second set of customized client data. The mechanism generates association rules for determining recommendations for preventing the set of predicted high-cost claimants from becoming high-cost claimants and applies the association rules to the second set of customized client data to generate a set of recommendations.
1 . A method, in a data processing system, for predictive and prescriptive analytics for managing high-cost claimants, the method comprising:
training a machine learning model using the set of de-identified claims data to predict high-cost claimants using the training data;
applying transfer learning by retraining the machine learning model using a first set of customized client data to generate a client-specific machine learning model;
applying the client-specific machine learning model to a second set of customized client data to identify a set of predicted high-cost claimants within the second set of customized client data;
generating association rules for determining recommendations for preventing the set of predicted high-cost claimants from becoming high-cost claimants; and
applying the association rules to the second set of customized client data to generate a set of recommendations.
2 . The method of claim 1 , wherein generating the association rules comprises:
finding frequent common features among the set of predicted high-cost claimants;
filtering the de-identified claims data for individuals having the frequent common features; and
applying association rule mining on the filtered de-identified claims data to generate a set of association rules, wherein each rule in the set of association rule associates a measure with an individual who is no longer a high-cost claimant.
3 . The method of claim 2 , wherein the measure comprises a procedure, a drug, or a rehabilitation measure.
4 . The method of claim 2 , wherein generating the association rules further comprises generating a confidence value for each measure and ranking the measures by confidence value.
5 . The method of claim 4 , wherein applying the association rules to the second set of customized client data comprises outputting a predetermined number of recommendations with the highest confidence value.
6 . The method of claim 1 , wherein generating the association rules comprises applying a Frequent Pattern (FP) Growth algorithm to the second set of customized client data.
7 . The method of claim 1 , wherein the machine learning model comprises a bidirectional Recurrent Neural Network with attention.
8 . A computer program product comprising a computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on a computing device, causes the computing device to:
train a machine learning model using the set of de-identified claims data to predict high-cost claimants using the training data;
apply transfer learning by retraining the machine learning model using a first set of customized client data to generate a client-specific machine learning model;
apply the client-specific machine learning model to a second set of customized client data to identify a set of predicted high-cost claimants within the second set of customized client data;
generate association rules for determining recommendations for preventing the set of predicted high-cost claimants from becoming high-cost claimants; and
apply the association rules to the second set of customized client data to generate a set of recommendations.
9 . The computer program product of claim 8 , wherein generating the association rules comprises:
finding frequent common features among the set of predicted high-cost claimants;
filtering the de-identified claims data for individuals having the frequent common features; and
applying association rule mining on the filtered de-identified claims data to generate a set of association rules, wherein each rule in the set of association rule associates a measure with an individual who is no longer a high-cost claimant.
10 . The computer program product of claim 9 , wherein the measure comprises a procedure, a drug, or a rehabilitation measure.
11 . The computer program product of claim 9 , wherein generating the association rules further comprises generating a confidence value for each measure and ranking the measures by confidence value.
12 . The computer program product of claim 11 , wherein applying the association rules to the second set of customized client data comprises outputting a predetermined number of recommendations with the highest confidence value.
13 . The computer program product of claim 8 , wherein generating the association rules comprises applying a Frequent Pattern (FP) Growth algorithm to the second set of customized client data.
14 . The computer program product of claim 8 , wherein the machine learning model comprises a bidirectional Recurrent Neural Network with attention.
15 . An apparatus comprising:
a processor; and
a memory coupled to the processor, wherein the memory comprises instructions which, when executed by the processor, cause the processor to:
train a machine learning model using the set of de-identified claims data to predict high-cost claimants using the training data;
apply transfer learning by retraining the machine learning model using a first set of customized client data to generate a client-specific machine learning model;
apply the client-specific machine learning model to a second set of customized client data to identify a set of predicted high-cost claimants within the second set of customized client data;
generate association rules for determining recommendations for preventing the set of predicted high-cost claimants from becoming high-cost claimants; and
apply the association rules to the second set of customized client data to generate a set of recommendations.
16 . The apparatus of claim 15 , wherein generating the association rules comprises:
finding frequent common features among the set of predicted high-cost claimants;
filtering the de-identified claims data for individuals having the frequent common features; and
applying association rule mining on the filtered de-identified claims data to generate a set of association rules, wherein each rule in the set of association rule associates a measure with an individual who is no longer a high-cost claimant.
17 . The apparatus of claim 16 , wherein the measure comprises a procedure, a drug, or a rehabilitation measure.
18 . The apparatus of claim 16 , wherein generating the association rules further comprises generating a confidence value for each measure and ranking the measures by confidence value.
19 . The apparatus of claim 15 , wherein generating the association rules comprises applying a Frequent Pattern (FP) Growth algorithm to the second set of customized client data.
20 . The apparatus of claim 15 , wherein the machine learning model comprises a bidirectional Recurrent Neural Network with attention.