Explainable artificial intelligence for genomic medicine

To improve the efficiency of investigatory work for experts in genomic medicine

To explain the reason and basis behind AI-generated findings in genomic medicine

Deep learning is one of the most representative technologies

in recent AI and shows high performance in pattern

recognition and analysis. However, as it cannot explain the

reasons for its judgment, it is called "black box AI."

There is a graph-structured data-based machine learning

technology called "Deep Tensor" that can directly analyse the

relations among numerous pieces of real-world data ranging

from intercompany transactions to material structures.

Additionally, there is also a technology for building a largescale knowledge base, which is called a "knowledge graph"

and consists of vast knowledge existing around the world

such as academic papers, by using our unique technology.

This technology identifies the factors (partial graphs) that

had a significant influence on an inference and coordinates

these with partial graphs from a knowledge graph, building

a series of pieces of information in the form of connections in

the knowledge graph as the basis for the findings.

People can combine these two technologies and develop a

system that enables AI to explain the reasons and basis

(evidence) for its judgment.

A use case of applying this explainable AI is genomic

medicine (for cancer treatment). The latest genomic

medicine helps detect patients' genetic defects that have

caused disease (cancer) and uses therapeutic drugs that

affect cancer cells produced by such genetic defects.

In genomic medicine today, a patient's normal and cancerous

cells are analysed with a next-generation sequencer; then, a

medical team uses the obtained genetic data to identify a

causal gene and determines the recommended treatment. It

takes at least two weeks for the medical team to conduct an

2

examination after completing genetic analysis. Unless the

cost and time problems are solved, spreading this

advantageous genomic medicine far and wide would be

difficult.

In this use case, the explainable AI trained Deep Tensor using

180 000 pieces of disease mutation data, successfully

embedding more than ten billion pieces of knowledge from

seventeen million medical articles and other materials into

a knowledge graph. Inputting genetic mutation data into this

system enables Deep Tensor to infer disease-causing factors

and enables the knowledge graph to find medical evidence to

justify the obtained results. Medical specialists then simply

are expected to review the flow of obtained inference logic,

thereby reducing the period between analysis and report

submission significantly from two weeks to a single day.