Predicting 30 day Readmission Probability For CHF Patients
A machine learning based solution is built to analyzes factors that caused readmission of CHF patients, computes readmission risk scores and suggest appropriate interventions. A patient clinical history 12 months prior to discharge and 30 day readmission records post discharge are used as required data. The solution makes use of K-means method and logistic regression model. The resulting model have ROC-AUC value of 0.65 and thus performed better than other known models in the healthcare space such as CMS HF model (AUC – 0.61) & CMS AMI model (AUC – 0.63)
Input variables : Patient’s demographics,
comorbid conditions (12 months prior history & 30 days post discharge details) data used
Output Variables : Factors that caused readmissions, compute readmission risk scores and suggest appropriate interventions
Metrics to Monitor
|
Statistical
|
:
|
Somers D |
Accuracy |
Precision and Recall |
Confusion Matrix |
F1 Score |
Roc and Auc |
Prevalence |
Detection Rate |
Balanced Accuracy |
Cohen's Kappa |
Concordance |
Gini Coefficent |
KS Statistic |
Youden's J Index
|
|
Business
|
:
|
Readmission Rate |
Avg Hospital LOS
|
|
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