Artificial intelligence-assisted digital therapy planning in precision oncology

Precision oncology has become part of routine patient care now. A growing number of cancer patients have molecular diagnostic test results of several hundred genes. More than 100 targeted or immunotherapies are registered today. Besides the new opportunities, oncologists may also face challenges, as multiple therapy options are available for the same biomarkers and the four-five co-occurring driver mutations present on avarage in the same tumor can alter the response to targeted therapies. Tumors often harbor multiple targets and biomarkers in parallel. Molecular tumor boards (MTBs) are responsible currently for setting the optimal treatment plan. However, multiple studies showed that different MTBs frequently choose different treatments for the same cases with complex molecular profiles. Standard decision making is required, so that we can measure and improve the efficacy of the decision making method itself. This way, the efficacy of decision making methods would become comparable, so patients could be treated based on the most effective of them. The digital drug assignment (DDA) algorithm provides an opportunity to standardize treatment decisions in cases where multiple targeted or immunotherapy options are available and the guidelines do not provide information on the difference of the clinical benefit regarding these compounds. The clinical benefit and safety of this method was demonstrated by re-analysis of the data of patients treated in the SHIVA01 clinical trial.