I built Bedside CRI and this biophysical sanctuary simulator to tackle an alarming clinical dilemma in lung cancer:
In EGFR-mutated non-small cell lung cancer (NSCLC), patients treated with 3 years of adjuvant osimertinib show remarkable survival (the ADAURA trial). However, longitudinal molecular residual disease monitoring reveals that ~68% of recurrences occur after the planned 3-year cessation, and ~80% of these relapsing patients harbor no secondary resistance mutations (such as EGFR C797S).
Instead of treating this as an unexplained genetic mystery, we modeled it from first physical principles: 1. Microvascular Perivascular Cords: Drug delivery from capillaries follows a cylindrical Modified Bessel function of the second kind K₀(r/λ). 2. Because of active tissue clearance and binding kinetics, the effective concentration drops below the 10 nM IC50 threshold well before reaching the hypoxic/necrotic boundary (r_ox ≈ 200 µm). This carves out a permanent, sub-therapeutic "pharmacological sanctuary" shell (Δr ≈ 34.1 µm, occupying ~31.3% of the viable cord volume). 3. In this sanctuary, persister cells survive in a cytostatic state during therapy. Once osimertinib is stopped at Month 36, exponential tumor regrowth kinetics (doubling time Ts ≈ 25 days) mathematically dictate clinical relapse within ~5.4 months.
Technical & Engineering details: - Zero external dependencies: The interactive bedside simulator runs client-side with pure HTML5 Canvas and vanilla JavaScript. - Modified Bessel K₀ integration is computed in real time at 60 fps directly in browser memory using numerical Simpson's rule. - The corresponding Python computational framework and CLI (cri-calc) are built exclusively with the Python standard library (no NumPy/SciPy/Pandas required for execution), guaranteeing permanent reproducibility.
Preprint (Springer Nature Research Square / indexed by Crossref): https://doi.org/10.21203/rs.3.rs-11530312/v1 GitHub Repository: https://github.com/Lukecele/spatial-continuum-oncology
I would love your feedback on the reaction-diffusion formulation, the simulator UI/UX, or the biophysical modeling!