Machine learning models in location based episode prediction
The disclosed embodiments include a method performed by server computer(s). The method includes obtaining private healthcare insurance claims data and public healthcare procedure code data from one or more data source(s), generating a model based the training data, and determining a multiplier for each of the many facilities based on the model. The model provides predictive results for variability between healthcare facilities thereby enabling consumer research in light of considerations in facility variability. The multiplier is indicative of a value used to scale the public healthcare procedure code data for a facility such that a healthcare service associated with the facility can be estimated based on the multiplier.
1. A method performed by one or more server computers to generate models of location based data for machine learning analysis, the method comprising:
obtaining, by a server, a plurality of healthcare claim data records corresponding to a plurality of healthcare facilities, each of the plurality of records having fields, the fields including:
a location;
a doctor;
a procedure code; and
an episode;
generating, by the server, a model based on training data including the plurality of records that predicts a likelihood for a given value in the doctor and the procedure code fields based on known values in the episode and location fields;
determining, by the server, a multiplier for each of the plurality of healthcare facilities based on the model, wherein a first multiplier associated with a first healthcare facility is indicative of a value used to scale a public healthcare fee schedule to a first procedure code associated with a first episode at the first healthcare facility performed by a first doctor;
receiving, by the server, user input specifying the episode and a region containing multiple values for the location field; and
determining, based on the user input and the model, corresponding procedure codes, corresponding doctors, and corresponding locations.
2. A method performed by one or more server computers, the method comprising:
obtaining private healthcare insurance claims data and public healthcare fee schedule data from one or more data sources;
generating a model based on training data including a combination of the private healthcare insurance claims data and the public healthcare fee schedule data; and
determining a multiplier for each of the plurality of healthcare facilities based on the model, wherein a first multiplier associated with a first healthcare facility is indicative of a value used to scale the public healthcare fee schedule for the first healthcare facility such that a cost for a healthcare service associated with the first healthcare facility is estimated based on the first multiplier of the healthcare facility.
3. The method of claim 2 , wherein the public healthcare fee schedule data is Centers for Medicare & Medicaid Services data.
4. The method of claim 2 , further comprising, prior to generating the model:
removing data identifying any patients from the private healthcare insurance claims data and the public healthcare fee schedule data.
5. The method of claim 2 , further comprising:
receiving a query to estimate a cost of a particular healthcare service performed at a particular healthcare facility;
calculating a particular cost estimate for the particular healthcare service based on a particular multiplier value associated with the particular healthcare facility; and
causing display of the cost estimate or data indicative of the cost estimate for the particular healthcare service.
6. The method of claim 5 , wherein the model defines a plurality of episodes of care, each having a frequency weighted combination of the particular healthcare service and at least one additional healthcare service, and the particular of cost estimate is based on a normalized aggregate of costs for the services including in the plurality of episodes of care.
7. The method of claim 5 , further comprising, prior to calculating the particular cost estimate:
receiving user input indicating a geographic region, wherein the query for the particular cost estimate is constrained by the geographic region such that the particular cost estimate is constrained to the geographic region.
8. The method of claim 5 , further comprising, prior to calculating the particular cost estimate:
receiving user input indicating a particular insurer, wherein the query for the cost estimate is constrained by the insurer such that the particular cost estimate is constrained to the insurer.
9. The method of claim 2 , wherein generating the model comprises:
ascertaining, from the private healthcare insurance claims data and the public healthcare fee schedule data, a plurality of data points each having healthcare facility and insurer pair; and
performing a Bayesian regression on the data points to generate the model as a linear model where the slope of the linear model represents a multiplier.
10. The method of claim 9 , further comprising:
obtaining additional data points;
performing a Bayesian regression including the additional data point to generate an updated model for an updated linear model.
11. The method of claim 2 , further comprising:
generating a plurality of additional models including a model that accounts for healthcare providers, a model that accounts for patient histories, a model that accounts for comorbidities, and a model that accounts for secondary healthcare providers.
12. The method of claim 2 , further comprising:
updating the model based on machine learning techniques.
13. The method of claim 2 , wherein the model facilitates any of discovering of re-negotiated costs by an insurer, projecting estimates of missing data in the private healthcare insurance claims data based on historical data, or identifying outlier data points of the model indicative of errors, fraudulent claims, negotiated cost patterns, or cost policies.
14. The method of claim 2 , further comprising:
receiving a query to estimate a cost of a particular healthcare service performed at a particular healthcare facility;
calculating a particular cost estimate for the particular healthcare service based on a particular multiplier value associated with the particular healthcare facility; and
causing display of a visualization indicative of the particular cost estimate.
15. The method of claim 2 , further comprising:
receiving a query to estimate a cost of a particular healthcare service performed at a particular healthcare facility;
calculating a particular cost estimate for the particular healthcare service based on a particular multiplier value associated with the particular healthcare facility; and
causing display of a visualization indicative of the particular cost estimate relative to a displayed map of a geographic region.
16. The method of claim 15 , wherein the particular cost estimate belongs to a plurality of cost estimates each associated with a visualization element indicating that the cost estimate as low for the geographic region, typical for the geographic region, or high for the geographic region.
17. The method of claim 15 , comprising:
causing display of a plurality of visualization elements associated with a plurality of cost estimates for the particular healthcare service, wherein each visualization element represents a different insurer;
causing display of information related to a cost estimate of a selected insurer.
18. The method of claim 15 , comprising:
causing display of a customizable cost estimate based on additional user input.
19. The method of claim 15 , wherein the visualization includes a range of cost estimates for an insurer as a function of a plurality of healthcare facilities, wherein the cost estimates are ranked and color coded to indicate a relative cost.
20. A computer system comprising:
a processor; and
memory including instructions that, when executed by the processor, cause the computer system to:
obtain private healthcare insurance claims data and public healthcare fee schedule data from one or more data sources;
generate a model based on training data including a combination of the private healthcare insurance claims data and the public healthcare fee schedule data; and
determine a multiplier for each of the plurality of healthcare facilities based on the model, wherein a multiplier associated with a healthcare facility is indicative of a value used to scale the public healthcare fee schedule for the healthcare facility such that a cost for a healthcare service associated with the healthcare facility is estimated based on the multiplier of the healthcare facility.
21. The system of claim 20 , the memory further comprising instructions that, when executed by the processor, cause the computer system to:
process receipt of a query to estimate a cost of a particular healthcare service performed at a particular healthcare facility;
calculate a particular cost estimate for the particular healthcare service based on a particular multiplier value associated with the particular healthcare facility; and
cause display of the cost estimate or data indicative of the cost estimate for the particular healthcare service.
22. The system of claim 21 , wherein the model defines a plurality of episodes of care, each having a frequency weighted combination of the particular healthcare service and at least one additional healthcare service, and the particular of cost estimate is based on a normalized aggregate of costs for the services including in the plurality of episodes of care.
23. The system of claim 21 , the memory further comprising instructions that, when executed by the processor prior to calculating the particular cost estimate, cause the computer system to:
receive user input indicating a geographic region, wherein the query for the particular cost estimate is constrained by the geographic region such that the particular cost estimate is constrained to the geographic region.
24. The system of claim 21 , the memory further comprising instructions that, when executed by the processor prior to calculating the particular cost estimate, cause the computer system to:
receive user input indicating a particular insurer, wherein the query for the cost estimate is constrained by the insurer such that the particular cost estimate is constrained to the insurer.
25. A non-transitory machine-readable storage medium storing instructions, an execution of which in a computer system causes the computer system to perform operations comprising:
obtaining private healthcare insurance claims data and public healthcare fee schedule data from one or more data sources;
generating a model based on training data including a combination of the private healthcare insurance claims data and the public healthcare fee schedule data; and
determining a multiplier for each of the plurality of healthcare facilities based on the model, wherein a multiplier associated with a healthcare facility is indicative of a value used to scale the public healthcare fee schedule for the healthcare facility such that a cost for a healthcare service associated with the healthcare facility is estimated based on the multiplier of the healthcare facility.