Drug delivery: Experiments, mathematical modelling and machine learning.

Area of Science:

• Pharmacology
• Computational Biology
• Machine Learning

Background:

• Accurate modeling of anticancer drug delivery and efficacy is crucial for effective therapy.
• Laboratory experiments often provide insufficient data for robust model parameter extraction.
• Challenges exist in identifying key physical model factors, such as drug-induced cell killing.

Purpose of the Study:

• To develop a reliable method for determining necessary data from laboratory experiments for anticancer drug modeling.
• To overcome data limitations by integrating real experiments, numerical simulations, and Machine Learning (ML).
• To create a predictive tool for drug delivery and efficacy by identifying physical model factors.

Main Methods:

• A hybrid approach combining real-world experiments, numerical simulations, and Artificial Neural Networks (ANN) based ML.
• Utilizing a mathematical-numerical model for tumor growth and drug delivery.
• Employing ANN-ML to integrate experimental results and provide feedback to the model for parameter identification.

Main Results:

• Successful development of a hybrid data-driven, physics-informed ML approach.
• Reliable identification of physical model factors, including the drug's killing action.
• Creation of a robust predictive tool for anticancer therapy modeling.

Conclusions:

• The integrated approach effectively addresses data scarcity in experimental drug delivery studies.
• This hybrid methodology enhances the predictive power of computational models in oncology.
• The physics-informed ML framework offers a promising direction for personalized cancer treatment strategies.