US-12619939-B2 - Systems and methods for determining unnecessary internal system utilization based on protocol adherence

Abstract

Systems and methods are disclosed for determining unnecessary internal system utilization based on protocol adherence. A method includes receiving a first data object, generating an entity data object, and generating a verified entity data object based on comparing one or more metrics of the entity data object against one or more predetermined threshold values, wherein entities of the verified entity data object are a subset of the entities of the entity data object. The method further includes generating a compliance indicator for each entity of the verified entity data object. The method furthermore includes generating a utilization adjustment data object and causing the utilization adjustment data object to be displayed on a Graphical User Interface (GUI).

Inventors

• Samara B. PRYWES
• James M. DOLSTAD
• Ian M. Smith
• Salina YIP
• Maxine GOLDSMITH
• Rajiv ARYA
• Yao Zhang

Assignees

• OPTUM, INC.

Dates

Publication Date

20260505

Application Date

20231229

Claims (19)

1. 1 . A computer-implemented method comprising: receiving, by one or more processors, a first data object, the first data object including: an entity data set including a plurality of entities; a performance history data set including a plurality of performance-related records for the plurality of entities; and one or more performance metric data sets for the plurality of entities, including active utilization data of the plurality of entities; generating, by the one or more processors, an entity data object for individual entities of the plurality of entities based on at least one of the entity data set, the performance history data set, or the one or more performance metric data sets; identifying, by the one or more processors and based on the entity data object for the individual entities of the plurality of entities, a subset of entities, of the plurality of entities, having a medication protocol for medical condition management that satisfies an eligibility criterion; determining, by the one or more processors and using a machine learning model, a risk score for the individual entities of the subset of entities based on the one or more performance metric data sets included in the first data object, wherein the machine learning model is trained, based on training data, to (i) identify associations between the active utilization data and a compliance risk for the individual entities of the subset of entities and (ii) output the risk score representing the compliance risk based on the identified associations; generating, by the one or more processors and based on the entity data object for the individual entities of the subset of entities, a compliance indicator indicating a likelihood of adherence of the respective individual entities of the subset of entities to the medication protocol; generating, by the one or more processors and based on the entity data object and the compliance indicator, an intervention data object for the respective individual entities of the subset of entities that includes a targeted alteration to the medication protocol; generating, by the one or more processors, a utilization adjustment data object based on the entity data object, the risk score, the compliance indicator, and the intervention data object for the respective individual entities of the subset of entities, wherein the respective individual entities of the subset of entities are mapped to one of a plurality of risk groups in the utilization adjustment data object; causing, by the one or more processors, the utilization adjustment data object to be displayed on a Graphical User Interface (GUI); in response to determining at least one individual entity of the subset of entities being mapped to a particular risk group of the plurality of risk groups, applying, by the one or more processors, the targeted alteration to the medication protocol for the at least one individual entity to cause administration of an altered medication protocol to the at least one individual entity, the altered medication protocol including the targeted alteration; monitoring, by the one or more processors, a performance of the machine learning model as a new data object is received; detecting, by the one or more processors and based on the monitoring, parameters associated with the new data object having a threshold difference from parameters associated with the training data that indicate a drift, the drift affecting the performance of the machine learning model; and in response to detecting the drift, applying, by the one or more processors, a correction mechanism to optimize the performance of the machine learning model, the correction mechanism including a retraining of the machine learning model based on the parameters associated with the new data object.
2. 2 . The computer-implemented method of claim 1 , wherein the active utilization data includes one or more active prescriptions of the medication protocol.
3. 3 . The computer-implemented method of claim 1 , wherein determining the risk score further comprises: weighting, by the one or more processors, one or more active utilizations included in the active utilization data for the individual entities of the subset of entities based on at least one of the one or more performance metric data sets, wherein a weight applied to each active utilization is determined based on a corresponding performance metric data set.
4. 4 . The computer-implemented method of claim 1 , wherein the medication protocol for the individual entities of the subset of entities includes one or more active prescriptions associated with each respective individual entity and generating the compliance indicator includes: receiving a complexity score for each active prescription of the one or more active prescriptions; assigning a total complexity score, the total complexity score representing a cumulative measure of the complexity score for each active prescription of the one or more active prescriptions; determining an adherence ratio using the total complexity score, the adherence ratio indicative of an expected proportion of days the respective individual entity is in compliance with an active prescription regimen of the medication protocol; comparing the adherence ratio against an adherence threshold; and generating the compliance indicator, wherein a first portion of the individual entities of the subset of entities meeting or surpassing the adherence threshold are marked as compliant, and a second portion of the individual entities of the subset of entities falling below the adherence threshold are marked as non-compliant.
5. 5 . The computer-implemented method of claim 1 , wherein identifying the subset of entities further comprises: identifying, by the one or more processors, the individual entities of the subset of entities as having at least one diagnostic indicator for a medical condition associated with the entity data object, the at least one diagnostic indicator being selected from a pre-determined plurality of diagnostic indicators.
6. 6 . The computer-implemented method of claim 5 , wherein the at least one diagnostic indicator is for a chronic medical condition.
7. 7 . The computer-implemented method of claim 1 , wherein generating the intervention data object for the respective individual entities of the subset of entities comprises: generating, by the one or more processors, the targeted alteration to the medication protocol to increase the likelihood of adherence to the medication protocol and reduce resource utilization by the respective individual entities of the subset of entities.
8. 8 . The computer-implemented method of claim 1 , wherein generating the intervention data object for the respective individual entities of the subset of entities comprises: determining a utilization offset metric associated with an expected reduction in resource utilizations by the respective individual entities in response to the targeted alteration to the medication protocol, wherein the utilization offset metric is included in the intervention data object.
9. 9 . The computer-implemented method of claim 1 , wherein the entity data object for the individual entities of the subset of entities includes information for one or more active prescriptions associated with the medication protocol, metrics indicating historical adherence to the medication protocol, and one or more external factors affecting adherence to the medication protocol.
10. 10 . The computer-implemented method of claim 1 , wherein the correction mechanism including the retraining of the machine learning model is a first correction mechanism applied when the detected drift is below a threshold.
11. 11 . The computer-implemented method of claim 10 , wherein when the detected drift is above the threshold, applying the correction mechanism further comprises: applying a second correction mechanism including one or more of incorporation of novel features or adjustment of hyperparameters of the machine learning model.
12. 12 . The computer-implemented method of claim 1 , wherein the retraining of the machine learning model includes one or more of modifying model parameters, adjusting weights, or recalibrating features.
13. 13 . A system comprising: one or more processors; and one or more non-transitory computer readable media storing processor-executable instructions which, when executed by the one or more processors, cause the one or more processors to perform operations comprising: receiving a first data object, the first data object including: an entity data set including a plurality of entities; a performance history data set including a plurality of performance-related records for the plurality of entities; and one or more performance metric data sets for the plurality of entities, including active utilization data of the plurality of entities; generating an entity data object for individual entities of the plurality of entities based on at least one of the entity data set, the performance history data set, or the one or more performance metric data sets; identifying a subset of entities, of the plurality of entities, having a medication protocol for medical condition management that satisfies an eligibility criterion; determining, using a machine learning model, a risk score for the individual entities of the subset of entities based on the one or more performance metric data sets included in the first data object, wherein the machine learning model is trained, based on training data, to (i) identify associations between the active utilization data and a compliance risk for the individual entities of the subset of entities and (ii) output the risk score representing the compliance risk based on the identified associations; generating, based on the entity data object for the individual entities of the subset of entities, a compliance indicator indicating a likelihood of adherence of the respective individual entities of the subset of entities to the medication protocol; generating, based on the entity data object and the compliance indicator, an intervention data object for the respective individual entities of the subset of entities that includes a targeted alteration to the medication protocol; generating a utilization adjustment data object based on the entity data object, the risk score, the compliance indicator, and the intervention data object for the respective individual entities of the subset of entities, wherein the respective individual entities of the subset of entities are mapped to one of a plurality of risk groups in the utilization adjustment data object; causing the utilization adjustment data object to be displayed on a Graphical User Interface (GUI); in response to determining at least one individual entity of the subset of entities being mapped to a particular risk group of the plurality of risk groups, applying the targeted alteration to the medication protocol for the at least one individual entity to cause administration of an altered medication protocol to the at least one individual entity, the altered medication protocol including the targeted alteration; monitoring a performance of the machine learning model as a new data object is received; detecting, based on the monitoring, parameters associated with the new data object having a threshold difference from parameters associated with the training data that indicate a drift, the drift affecting the performance of the machine learning model; and in response to detecting the drift, applying a correction mechanism to optimize the performance of the machine learning model, the correction mechanism including a retraining of the machine learning model based on the parameters associated with the new data object.
14. 14 . The system of claim 13 , wherein the active utilization data includes one or more active prescriptions of the medication protocol.
15. 15 . The system of claim 13 , wherein determining the risk score further comprises: applying a weight to one or more active utilizations included in the active utilization data for the individual entities of the subset of entities based on at least one of the one or more performance metric data sets, wherein a weight applied to each active utilization is determined based on a corresponding performance metric data set.
16. 16 . The system of claim 13 , wherein the medication protocol for the individual entities of the subset of entities includes one or more active prescriptions associated with each respective individual entity and generating the compliance indicator includes: receiving a complexity score for each active prescription of the one or more active prescriptions; assigning a total complexity score, the total complexity score representing a cumulative measure of the complexity score for each active prescription of the one or more active prescriptions; determining an adherence ratio using the total complexity score, the adherence ratio indicative of an expected proportion of days the respective individual entity is in compliance with an active prescription regimen of the medication protocol; comparing the adherence ratio against an adherence threshold; and generating the compliance indicator, wherein a first portion of the individual entities of the subset of entities meeting or surpassing the adherence threshold are marked as compliant, and a second portion of the individual entities of the subset of entities falling below the adherence threshold are marked as non-compliant.
17. 17 . The system of claim 13 , wherein identifying the subset of entities further comprises: identifying the individual entities of the subset of entities as having at least one diagnostic indicator for a medical condition associated with the entity data object, the at least one diagnostic indicator being selected from a pre-determined plurality of diagnostic indicators associated with a chronic medical condition.
18. 18 . One or more non-transitory computer-readable media storing instructions that, when executed by one or more processors, cause the one or more processors to perform operations comprising: receiving a first data object, the first data object including: an entity data set including a plurality of entities; a performance history data set including a plurality of performance-related records for the plurality of entities; and one or more performance metric data sets for the plurality of entities, including active utilization data of the plurality of entities; generating an entity data object for individual entities of the plurality of entities based on at least one of the entity data set, the performance history data set, or the one or more performance metric data sets; identifying, based on the entity data object for the individual entities of the plurality of entities, a subset of entities, of the plurality of entities, having a medication protocol for medical condition management that satisfies an eligibility criterion; determining, by the one or more processors and using a machine learning model, a risk score for the individual entities of the subset of entities based on the one or more performance metric data sets included in the first data object, wherein the machine learning model is trained, based on training data, to (i) identify associations between the active utilization data and a compliance risk for the individual entities of the subset of entities and (ii) output the risk score representing the compliance risk based on the identified associations; generating, based on the entity data object for the individual entities of the plurality of entities, a compliance indicator indicating a likelihood of adherence of the respective individual entities of the subset of entities to the medication protocol; generating, based on the entity data object and the compliance indicator, an intervention data object for the respective individual entities of the subset of entities that includes a targeted alteration to the medication protocol; generating a utilization adjustment data object based on the entity data object, the risk score, the compliance indicator, and the intervention data object for the respective individual entities of the subset of entities, wherein the respective individual entities of the subset of entities are mapped to one of a plurality of risk groups in the utilization adjustment data object; causing the utilization adjustment data object to be displayed on a Graphical User Interface (GUI); in response to determining at least one individual entity of the subset of entities being mapped to a particular risk group of the plurality of risk groups, applying the targeted alteration to the medication protocol for the at least one individual entity to cause administration of an altered medication protocol to the at least one individual entity, the altered medication protocol including the targeted alteration; monitoring a performance of the machine learning model as a new data object is received; detecting, based on the monitoring, parameters associated with the new data object having a threshold difference from parameters associated with the training data that indicate a drift, the drift affecting the performance of the machine learning model; and in response to detecting the drift, applying a correction mechanism to optimize the performance of the machine learning model, the correction mechanism including a retraining of the machine learning model based on the parameters associated with the new data object.
19. 19 . The one or more non-transitory computer-readable media of claim 18 , wherein the medication protocol for the individual entities of the subset of entities includes one or more active prescriptions associated with each respective individual entity and generating the compliance indicator includes: receiving a complexity score for each active prescription of the one or more active prescriptions; assigning a total complexity score, the total complexity score representing a cumulative measure of the complexity score for each active prescription of the one or more active prescriptions; determining an adherence ratio using the total complexity score, the adherence ratio indicative of an expected proportion of days the respective individual entity is in compliance with an active prescription regimen of the medication protocol; comparing the adherence ratio against an adherence threshold; and generating the compliance indicator, wherein a first portion of the individual entities of the subset of entities meeting or surpassing the adherence threshold are marked as compliant, and a second portion of the individual entities of the subset of entities falling below the adherence threshold are marked as non-compliant.

Description

TECHNICAL FIELD The present disclosure generally relates to the field of data analytics. In particular, the present disclosure relates to systems and methods for modeling protocol complexity based on analyzing various data sources and predicting the level of protocol adherence to generate interventions for increased resource utilization efficiency. BACKGROUND Ambulatory Care Sensitive Conditions (ACSCs) are conditions that can be effectively managed in external settings, preventing the need for internal system utilization. Adherence to recommended protocols such as, e.g., medication regimen, is a crucial element in the successful management of these conditions. However, complex protocols can present a significant barrier to optimal adherence. Existing strategies to manage this issue include protocol reconciliation, entity education, reminder systems, and simplification of protocols whenever possible. However, these techniques suffer from one or more issues and may be improved in one or more ways. For instance, current techniques often struggle to identify and respond to the individual factors that contribute to an entity's ability to adhere to a complex protocol. Protocol reconciliation is a useful tool, but it primarily focuses on ensuring correct prescription and usage, rather than simplifying the protocol itself. Entity education initiatives are crucial, yet they may not fully address the challenges posed by complex protocols, nor are they sufficiently personalized. Reminder systems can be beneficial but rely heavily on entity engagement and technological capabilities. The process of simplifying protocols is often reactive, rather than proactive, and may not adequately consider the entity's unique circumstances and capabilities. The consequence of these shortcomings includes suboptimal protocol adherence, which can lead to poorer management of entity's conditions and increased internal system and/or emergency resource utilization. Therefore, there is a need for a more sophisticated and predictive approach to managing protocol adherence in the context of, for example, ACSC management. This disclosure is directed to addressing the above-mentioned challenges. The background description provided herein is for the purpose of generally presenting the context of the disclosure. Unless otherwise indicated herein, the materials described in this section are not prior art to the claims in this application and are not admitted to be prior art, or suggestions of the prior art, by inclusion in this section. SUMMARY The present disclosure addresses the technical problem(s) described above or elsewhere in the present disclosure and improves the state of conventional healthcare management techniques. In some aspects, the techniques described herein relate to a computer-implemented method including: receiving, by one or more processors, a first data object, the first data object including: an entity data set containing a plurality of entities; a performance history data set containing a plurality of performance-related records; an event data set; and one or more performance metric data sets; generating, by the one or more processors, an entity data object based on at least one of the entity data set, the performance history data set, or the one or more performance metric data sets; generating, by the one or more processors, a verified entity data object based on comparing one or more metrics of the entity data object against one or more predetermined threshold values, wherein entities of the verified entity data object are a subset of the entities of the entity data object; generating, by the one or more processors, a compliance indicator for each entity of the verified entity data object; generating, by the one or more processors, a utilization adjustment data object based on the verified entity data object, a risk score associated with one or more performance metric data sets, and the compliance indicator for each entity; and causing, by the one or more processors, the utilization adjustment data object to be displayed on a Graphical User Interface (GUI). In some aspects, the techniques described herein relate to a system including memory and one or more processors communicatively coupled to the memory, the one or more processors configured to: receive a first data object, the first data object including: an entity data set containing a plurality of entities; a performance history data set containing a plurality of performance-related records; an event data set; and one or more performance metric data sets; generate an entity data object based on at least one of the entity data set, the performance history data set, or the one or more performance metric data sets; generate a verified entity data object based on comparing one or more metrics of the entity data object against one or more predetermined threshold values, wherein entities of the verified entity data object are a subset of the entities of the entity data obj