cesarmusic10/Risk-Mechanism-Theory-Index-RMTI-

Reference implementation of the Risk Mechanism Theory Index (RMTI): shared equations and framework, with applied modules for oncology risk and disaster/climate risk (Sub-Saharan Africa cohort) # Risk Mechanism Theory Index (RMTI) Reference implementation of the **Risk Mechanism Theory Index (RMTI)**, Marolla (2025), *"RMTI: A Systemic Risk Mechanism Design and Theory Framework."* This repository holds the shared mathematical core once, and applies it across domains: oncology risk and disaster/climate risk, with more applications to follow. ## Framework RMTI scores residual risk for an asset, patient, or cohort using a fixed indicator backbone (Table 9.4): | Symbol | Meaning | Effect on R | |--------|---------------------------------------------|------------------| | E | Exposure (RMTre) | raises R | | V | Vulnerability (RMTseen) | raises R | | rho | Resilience / treatment capacity (RMTrl) | lowers R, via (1-rho) | Core equations: - **Eq. (1)/(2):** `IR = L * E * V`, `R = IR * (1 - rho)` - **Eq. (3):** `R = sum_h,c w_c * L_h * E_h,c * V_h,c * (1 - rho_h,c)` (multi-factor/multi-category generalization) - **Eq. (4):** advisory log-normalized exposure baseline from population/cohort size N - **Eq. (5)/(6):** Expected Annual Loss, `EAL = L * C * (1 - rho)`, and its multi-factor generalization - **Eq. (7):** per-capita / per-patient equity rate Risk scores map to five fixed, pre-registered tiers (Table 3: Low, Moderate, High, Severe, Critical) and consequence severity uses a fixed five-level scale (Table 2). These tables are never re-derived per sample — that fixity is what lets an R score mean the same thing across every application. ## Repository structure ``` rmti-framework/ ├── core/ │ └── rmti_core.py # shared equations, tables, validated core, multi-factor extension ├── oncology/ │ └── ... # HCC / oncology application (see oncology/README.md) ├── disaster_climate/ │ └── rmti_ssa.py # disaster/climate application: six-city Sub-Saharan Africa cohort └── docs/ └── ... # supp …