Patient treatment resource utilization predictor

1. A computer implemented method comprising:

obtaining a schema defining variables representing patient treatment experiences for one or more health conditions;

obtaining data from historical patient treatment experiences, the data being arranged in accordance with the schema;

training a machine learning system based on a training set of the data for a specified time-period;

generating a model from the trained machine learning system, the model being based on variables selected by the trained machine learning system to predict patient treatment experiences for the one or more health conditions;

testing the trained machine learning system on a set of test data from the obtained data from historical patient treatment experiences;

classifying test data patient treatment experiences via the trained machine learning system;

identifying the number of patient treatment experiences misclassified by the machine learning system as “in story”;

identifying the number of patient treatment experiences correctly classified by the machine learning system as “out of story”; and

generating a new model using one or more qualifiers based on incorrectly classified in story patient treatment decisions to increase the accuracy of the machine learning system.

2. The method of claim 1 , wherein the patient treatment experiences include at least one of a length of stay corresponding to a patient treatment procedure, readmittance following a hospital stay, outpatient visits turning into a hospitalization, emergency room visits frequency, complication rate, expenditure, population emergency room visit rate, population hospital admission rate.

3. The method of claim 2 , wherein the data corresponds to a single health care facility and wherein the specified time-period comprises a time-period starting from a time following a change in procedures at the single health care facility.

4. The method of claim 3 , and further comprising:

generating a first statistic for the patient treatment experience for a first time-period prior to the change in procedures;

generating a second statistic for the patient treatment experience for a second time-period following the change in procedures; and

determining the effectiveness of the change in procedures as a function of the patient treatment experience.

5. The method of claim 4 , wherein the length of stay statistic comprises a percent of patient treatment experiences that are above a selected threshold over an average length of stay for a respective time-period.

6. The method of claim 1 , wherein the model comprises a decision tree including multiple rules and further comprising pruning the decision tree as a function of attribute usage in the decision tree.

7. The method of claim 1 , wherein the model comprises a decision tree including multiple rules and further comprising boosting the decision tree by generating multiple different decision trees from the trained machine learning system and combining results from the multiple different decision trees to predict patient treatment experiences for the one or more health conditions.

8. The method of claim 1 , wherein the data is stored in a database in accordance with a schema defining the data comprising:

a dependent variable name and corresponding discrete value selected from a set of discrete values;

a first set of multiple independent variables having corresponding discrete values; and

at least one independent variable having a corresponding continuous value.

9. A computer implemented method comprising:

accessing a trained machine learning system that has been trained by:

obtaining a schema defining variables representing patient treatment experiences for one or more health conditions;

obtaining data from historical patient treatment experiences, the data being arranged in accordance with the schema;

training a machine learning system based on a training set of the data for a specified time-period;

generating a model from the trained machine learning system, the model being based on variables selected by the trained machine learning system to predict patient treatment experiences for the one or more health conditions;

testing the trained machine learning system on a set of test data from the obtained data from historical patient treatment experiences;

classifying test data patient treatment experiences via the trained machine learning system;

identifying the number of patient treatment experiences misclassified by the machine learning system as “in story”;

identifying the number of patient treatment experiences correctly classified by the machine learning system as “out of story”;

generating a new model using one or more qualifiers based on incorrectly classified in story patient treatment decisions to increase the accuracy of the machine learning system; and

analyzing, using the trained machine learning system, the patient discharge records to measure the performance of a health care facility.

10. The computer implemented method of claim 9 , wherein measuring the performance of the health care facility further includes identifying one or more aspects of the provided health care that is above a baseline or identifying one or more aspects of the provided health care that is below a baseline.

11. The method of claim 9 , wherein the patient treatment experiences include at least one of a length of stay corresponding to a patient treatment procedure, readmittance following a hospital stay, outpatient visits turning into a hospitalization, emergency room visits frequency, complication rate, expenditure, population emergency room visit rate, population hospital admission rate.

12. The method of claim 11 , wherein the data corresponds to a single health care facility and wherein the specified time-period comprises a time-period starting from a time following a change in procedures at the single health care facility.

13. The method of claim 12 and further comprising:

generating a first statistic for the patient treatment experience for a first time-period prior to the change in procedures;

generating a second statistic for the patient treatment experience for a second time-period following the change in procedures; and

determining the effectiveness of the change in procedures as a function of the patient treatment experience.

14. The method of claim 13 , wherein the length of stay statistic comprises a percent of patient treatment experiences that are above a selected threshold over an average length of stay for a respective time-period.

15. The method of claim 9 , wherein the model comprises a decision tree including multiple rules and further comprising pruning the decision tree as a function of attribute usage in the decision tree.

16. The method of claim 9 , wherein the model comprises a decision tree including multiple rules and further comprising boosting the decision tree by generating multiple different decision trees from the trained machine learning system and combining results from the multiple different decision trees to predict patient treatment experiences for the one or more health conditions.

17. The method of claim 9 , wherein the data is stored in a database in accordance with a schema defining the data comprising:

a dependent variable name and corresponding discrete value selected from a set of discrete values;

a first set of multiple independent variables having corresponding discrete values; and

at least one independent variable having a corresponding continuous value.

18. A system comprising:

a processor and a memory device coupled to the processor and having instructions stored thereon for execution by the processor to perform operations comprising:

accessing a trained machine learning system that has been trained by:

obtaining a schema defining variables representing patient treatment experiences for one or more health conditions;

obtaining data from historical patient treatment experiences, the data being arranged in accordance with the schema;

training a machine learning system based on a training set of the data for a specified time-period;

generating a model from the trained machine learning system, the model being based on variables selected by the trained machine learning system to predict patient treatment experiences for the one or more health conditions;

testing the trained machine learning system on a set of test data from the obtained data from historical patient treatment experiences;

classifying test data patient treatment experiences via the trained machine learning system;

identifying the number of patient treatment experiences misclassified by the machine learning system as “in story”;

identifying the number of patient treatment experiences correctly classified by the machine learning system as “out of story”;

generating a new model using one or more qualifiers based on incorrectly classified in story patient treatment decisions to increase the accuracy of the machine learning system; and

analyzing, using the trained machine learning system, the patient discharge records to measure the performance of a health care facility.

19. The computer implemented method of claim 18 , wherein measuring the performance of the health care facility further includes identifying one or more aspects of the provided health care that is above a baseline or identifying one or more aspects of the provided health care that is below a baseline.

20. The method of claim 18 , wherein the patient treatment experiences include at least one of a length of stay corresponding to a patient treatment procedure, readmittance following a hospital stay, outpatient visits turning into a hospitalization, emergency room visits frequency, complication rate, expenditure, population emergency room visit rate, population hospital admission rate.