Japanese company Fujitsu in collaboration with the Institute of Medical Sciences at the University of Tokyo have demonstrated how artificial intelligence (AI) can improve the efficiency of treatment planning in genomic cancer medicine.
In the field of genomic cancer medicine, creating treatment plans derived from genomic information remains a costly and time-consuming process.
The newly developed technology allows for the generation of a knowledge graph of cancer genomic medicine, which could be used to create treatment plans that are able to take into account the effects of an ongoing treatment.
Verification test experiments using AI have enabled the Department of Haematology and Oncology at the University of Tokyo Institute of Medical Sciences to reduce the amount of work required to determine a treatment for acute myeloid leukaemia by more than half, meaning improved efficiency.
As research continues, Fujitsu laboratories will support the work of doctors by expanding artificial intelligence technology to address treatment plans for a range of different types of cancer, and hopefully contribute to the overall advancement of genomic cancer medicine.
The goal of genomic cancer medicine is to provide optimal medical care for each patient by identifying genomic mutations in cancer patients and predicting the likelihood of disease, as well as how a patient will respond to medication - including side effects. Since June 2019 cancer gene panel tests have been covered by health insurance in Japan, and industry experts anticipate that demand for these tests will increase.
The genomic cancer medicine remains a complex process to successfully achieve. It is still necessary for medical specialists to thoroughly search through articles one by one from a database to establish appropriate treatment methods, along with further research to determine the likely effects of treatment on the patient.
With artificial intelligence, a knowledge database on the relationship between genetic mutations and therapeutic drugs, and on the relationship between therapeutic drugs and their effects, can be automatically generated from medical research. Fujitsu's AI language processing technology is able to identify relevant words and phrases from medical research papers to speed up the treatment process. With this technology, 2.4 million elemental relationships from 860,000 medical documents are automatically extracted as ‘knowledge’, in order to build a database of graphs for appropriate genomic cancer medicine.
The joint study with Fujitsu and the University of Tokyo measures the time required for four physicians specialising in haematological malignancies at Tokyo’s Institute of Medical Sciences to search and examine documents, using technology based on previous cases of acute myeloid leukaemia. It also evaluates the efficiency of the treatment process comparing the use of the newly developed technology and without it.
The study showed that the new AI technology reduced treatment planning time significantly.
It is currently estimated around 12,000 people are diagnosed with leukaemia each year in Japan. If genomic medical treatments are administered to all patients, Fujitsu’s new technology could reduce examination processing time from approximately 6,000 hours to around 3,000 hours, or perhaps even less. This means the whole processing time of determining the appropriate treatment for blood cancer patients could be rapidly reduced with the support of AI.