Tailoring combinational therapy with Monte Carlo method-based regression modeling

Area of Science:

• Computational Biology
• Pharmacology
• Infectious Diseases

Background:

• Combinatorial drug therapies offer greater efficacy than monotherapies for viral infections.
• Optimizing drug doses is crucial for maximizing therapeutic effects and minimizing adverse events.
• Existing methods for accelerating drug combination optimization are hindered by biological assay noise and poor reproducibility.

Purpose of the Study:

• To develop a novel algorithm for rapid and robust identification of effective drug combinations.
• To address limitations of conventional methods in handling experimental noise and improving optimization efficiency.
• To enable personalized drug combination therapies for viral diseases.

Main Methods:

• Introduced the Regression Modeling Enabled by Monte Carlo Method (ReMEMC) algorithm.
• ReMEMC transforms sample variations into probability distributions for regression coefficients and predictions.
• Validated ReMEMC through in silico simulations and experimental application to COVID-19.

Main Results:

• ReMEMC demonstrated superior robustness and performance compared to conventional regression methods in simulations.
• Successfully identified an optimal 3-drug combination for COVID-19 within two experimental rounds.
• The identified optimal combination achieved significant viral load reduction compared to non-optimized combinations and monotherapy.

Conclusions:

• ReMEMC is an efficient and universal tool for accelerating dose combination optimization.
• The algorithm facilitates rapid identification of effective combinational therapies, including personalized strategies.
• This approach holds promise for improving treatment outcomes in viral infections.