Worldwide efforts are being made to reduce the toxicity of various therapies for the better management of cancer in patient populations and improve the quality of their lives. Traditional treatment regimens that combine various radiation therapies and multiple drug doses are largely focused on reducing the size of tumors. Oncologists and medical professionals usually resort to administering maximum safe doses of the therapies. This has frequently caused side-effects leading to poor responses and unexpectedly low outcomes in patients. One good approach that has shown potential is employing artificial intelligence (AI) models in minimizing the toxicity of chemotherapy and radiotherapy in an aggressive malignant brain tumor while ensuring that these therapies are still efficacious enough to shrink the tumor size.
Researchers at Massachusetts Institute of Technology (MIT) using innovative machine-learning techniques were successful in finding optimal dosage regimens for several patients. The regimens are characterized by dosage with lowest possible potency and frequency of drug doses and maximum effect on reducing the tumor.
Iterations based on Machine Learning Techniques learns about Parameters to develop New Treatment Regimens
The researchers employed reinforced learning (RL) technique to train their AI models by treating 50 simulated glioblastoma patients. These patients were taken from a large database and as many as 20,000 trial-and-error tests were conducted to learn about the parameters. The model essentially used the principle of iteration to look at the toxicity and efficacy of each traditional treatment regimen at a set dosing interval—weeks or months—and arrived at an optimal size of the dosage. They then tested their AI model on new set of 50 patients and compared the results with those of a traditional regimen that combined drugs temozolomide (TMZ) and vincristine (PVC).
AI helps in finding Model Precision Medicine-Based Treatments for Glioblastoma Patients
The AI model that researchers tested considered patients individually as well as in a single cohort group. Traditional treatment regimens, argue the researchers, do not give due consideration to factors such as tumor size, genetic profiles, biomarkers, and medical histories of the patients. Even the clinical trials conducted so far have ignored the effect of these parameters while deciding on the dosage.
The AI model scientists developed explored the dynamic of the potential adverse consequences of dosage and tumor reduction. Ultimately they succeeded at finding out a perfect balance—known as the optimal treatment regimen that changes with each patient. The regimen is the framework for developing precision medicine-based treatments for the most common brain tumor in adults.