# Validation of an agent-based model for cell interactions in a microfluidic chip

## TL;DR

This paper validates a computational model that simulates cell interactions in a microfluidic chip, improving predictions of cell behavior for drug testing.

## Contribution

The novel contribution is the integration of agent-based modeling with ABC-SMC for parameter estimation in microfluidic cell co-culture systems.

## Key findings

- The model accurately reproduces observed cellular behavior.
- It distinguishes between different experimental conditions effectively.

## Abstract

Objectives: Microfluidic cell Co-Culture, Tissue Co-Culture and Organ-on-Chip (OoC) technologies enable modeling of tissues and organs in vitro, facilitating cell-environment interaction studies and early therapeutic evaluation. The combination of physiology-based models, agent-based models (ABMs), cellular automata, and in-vitro modelling of complex processes provides a powerful tool to formalize, quantify, and predict observed phenomena.

Methods: Estimating parameters for these hybrid computational models using observational data is challenging. Approximate Bayesian computation (ABC) is particularly well suited for this task due to the intractability of the likelihood function. This work extends a hybrid ABM for a cell co-culture experiment on a chip. Cell tracking data is used to estimate model parameters via a Sequential Monte Carlo ABC (ABC-SMC) approach.

Results: The resulting model accurately reproduces observed cellular behavior and distinguishes between different experimental conditions.

Conclusion: The combination of cell co-culture and microfluidic technology with hybrid computational models and ABC-SMC provides a robust framework for modeling and predicting cellular behavior in vitro, enhancing the potential for early therapeutic evaluation and understanding of cell-environment interactions.