Compliance Rate Projection
CMS STAR ratings are calculated and assigned based on the Contract’s performance for a set of measures. These performance scores/Compliance ratios are calculated at a measure level, and are based on how the contracts performs for that respective measure in the given year. Further, they are assigned equivalent star ratings, and a weighted average of all the measure level ratings give the final STAR rating. It is executed yearly, works on HEDIS, PDE, CHAPS, HOS and Operational DataSource.
Methodology:
• A regression-based model (SVM/ Light GBM) is used to predict the projection of next year’s un-scaled measure-level score
• We build one model per measure. So, for any given year if there are 45-part C and D measures, SDE will have 45 different measure level regression models
• Data Source: CMS report cards
• Predictions are made for all measure-contract combinations, wherever applicable
• Result is measured using MAE
Input variables : CMS Report Card - Information derived from past years performance score for the given measure (e.g. -Average score, data deviation, etc.), and contract level enrolment related information
Output Variables : Predict the projection of next year’s un-scaled measure-level score
Metrics to Monitor
|
Statistical
|
:
|
Mallow's CP |
R Squared |
Mean Square Error |
Adjusted R Squared |
Mean Absolute Error |
Huber Loss
|
|
Infrastructure
|
:
|
Log Bytes |
Logging/User/IAMPolicy |
Logging/User/VPN |
CPU Utilization |
Memory Usage |
Error Count |
Prediction Count |
Prediction Latencies |
Private Endpoint Prediction Latencies |
Private Endpoint Response Count
|
Visit Model :
medictiv.citiustech.com