Fig 1.
Flowchart of analysis for survival path mapping.
(A) The time-series data of participants were converted into data of time slices with constant time interval. (B) Starting from the first time slice, data of the whole population would initially undergo the processing cycle (PC) for feature selection and subgroup subdivision, followed by PC in each subdivided subgroup data at the next time slice. The subdivision of parent node in the No. (n-1) time slice determine the population of the nodes in No.n time slice. The program continue until user defined last time slice.
Fig 2.
The data structure needed to construct survival path model. The columns includes time points, observational variables, treatment variables (optional) and outcome variables. Each row represent observation data at specific time point, and the date is sorted in ascending order for each participant/patient.
Fig 3.
Graph of survival path computed as demonstration.
The graph was built using ggtree based on tree-structure data.
Fig 4.
Demonstration of survival curves in comparing survival between different nodes (a) and subgroups with different treatment arms at specific nodes (b).