Grouping of patients based on similar disease progression pathways, also known as Patient Subtyping is extremely important to account heterogeneity in the patients which might lead providing treatment to patients considering their unique health status. An approach call Time-Aware LSTM (T-LSTM) is used to handle irregular time intervals in longitudinal patient records and clustering clustering of patients. A standard LSTM algorithm can handle long term dependencies for the inputs but it assumes uniformly distributed elapsed time between elements of a sequence. Which is not the case in most of the patient records. Hence to account the time irregularity in the data T-LSTM is proposed. In it the dependency on previous record is considered with the time duration between current and previous record. For application, T-LSTM is used on Parkinson's Progression Markers Initiative (PPMI) Data to identify biomarkers of the progression of Parkinson's disease and to predict the target sequence for patients.
Input variables : Patient Characteristics
Output Variables : Progression of Parkinson's Disease
| 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 | : | Population at High Risk of Disease | Risk by Geography | Risk by Demographics | Risk by Clinical Parameters |
| 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 : github.com
Additional links : biometrics.cse.msu.edu
| Model Category | : | Public |
| Date Published | : | August, 2017 |
| Healthcare Domain | : | Provider |
| Code | : | github.com |
| Health Risk Management |
| Risk Progression |