A groundbreaking study, recently published in Nature Biomedical Engineering by a collaborative team from Massachusetts Institute of Technology (MIT), Brigham and Women’s Hospital at Harvard Medical School, Duke University, and their research affiliates, has shed light on the potential of AI in predicting which drugs might compromise the efficacy of others.

The prevalence of prescription medications, particularly those targeting mental health, pain management, and hypertension, has seen a dramatic surge in recent years. According to Statista’s 2021 data, the total number of drug prescriptions in the U.S. has skyrocketed from 3.95 billion in 2009 to an estimated 6.7 billion in 2022, with projections indicating a further increase to 4.38 billion by 2025. With a significant portion of the population relying on multiple prescriptions concurrently, the risk of drug interactions looms large, underscoring the critical need for predictive tools in healthcare.

Central to the study’s methodology was the identification of drug transporters, membrane proteins that facilitate the passage of medications across cellular barriers. By focusing on three key drug transporters crucial for oral medication absorption—breast cancer resistance protein (BCRP), multidrug resistance-associated protein 2 (MRP2), and P-glycoprotein (Pgp)—the researchers aimed to elucidate the intricate interplay between various drugs within the gastrointestinal tract.

Employing a sophisticated tissue culture system derived from the small intestines of laboratory pigs, the researchers leveraged small interfering ribonucleic acid (siRNA) to modulate the expression of drug transporters, thereby unraveling their roles in drug absorption. This innovative approach enabled the identification of over 20 drug transporters, serving as the foundation for training an AI machine-learning model capable of predicting potential drug interactions.

Remarkably, the AI model successfully predicted previously unknown drug-transporter interactions for a diverse array of clinical and investigational drugs. Subsequent validation against pharmacologic data from hospital patients confirmed the model’s accuracy in foreseeing drug interactions.

The implications of this research are profound. By providing invaluable insights into potential drug interactions, the AI model stands to revolutionize drug development, streamlining the process and mitigating risks associated with adverse drug reactions.

As the boundaries of AI and machine learning continue to expand, the prospects for improved patient safety and therapeutic efficacy in the realm of pharmaceuticals are brighter than ever before.

Article written by Cammy Rosso| Image by Unsplash



Psicology Today