Empowering photodynamic therapy with artificial intelligence: current trends and future directions

Area of Science:

• Biomedical Engineering
• Photomedicine
• Artificial Intelligence in Medicine

Background:

• Photodynamic therapy (PDT) utilizes photosensitizers, light, and oxygen to generate reactive oxygen species for targeted cell destruction, showing efficacy in various cancers and non-malignant conditions.
• Personalizing PDT is challenging due to patient-specific variations in tissue optics, photosensitizer pharmacokinetics, and tumor heterogeneity.
• Artificial intelligence (AI), encompassing machine learning and deep learning, presents a powerful approach to overcome these personalization challenges through data-driven optimization.

Purpose of the Study:

• To review the integration of AI across the entire photodynamic therapy pipeline, from drug development to clinical decision-making.
• To highlight AI-driven strategies for enhancing photosensitizer design, nanoparticle drug delivery, and treatment planning.
• To critically assess the current limitations and future directions of AI in advancing PDT.

Main Methods:

• Analysis of AI applications in quantitative structure-activity relationship modeling, graph neural networks, and generative models for photosensitizer development.
• Examination of machine learning for optimizing nanoparticle synthesis, predicting nano-bio interactions, and modeling drug release.
• Review of AI in treatment planning via real-time optical property estimation and clinical decision-making through outcome prediction using imaging data.

Main Results:

• AI facilitates the development of novel photosensitizers and sophisticated nanoparticle delivery systems for PDT.
• Machine learning models enhance treatment planning accuracy and enable real-time monitoring of therapeutic response.
• AI integration addresses key challenges in PDT personalization, including tissue heterogeneity and pharmacokinetic variability.

Conclusions:

• AI offers transformative potential for optimizing photodynamic therapy across its pipeline, leading to more personalized and effective treatments.
• Addressing challenges like small datasets and model interpretability is crucial for successful clinical translation of AI in PDT.
• Future directions include federated learning, explainable AI, and regulatory frameworks to fully realize AI's impact on PDT patient outcomes.