Displaying present location in the site.
AI Initiative of a regional city
Report from Edmonton City, Canada(1/2)Utilization of big data in the health carePrediction of occurrence of individual’s disease and its prevention
This article is written by Mr. Kenji Kobayashi, Senior Fellow, Institute for International Socio-Economic Studies（NEC Group）
Canada is lately attracting global attention as “Hidden AI Power”. In the year 2017 the government announced the first in the world the national strategy for AI, and is striving to enhance domestic three cities, namely Toronto, Montreal and Edmonton, where globally acknowledged AI Research Institutes are located, not only for research and study but also for Industrial “AI Hub”.
In this issue I would like to report the challenge and arrangement of Edmonton City which might not be so popular nor familiar in Japan. (Reports in two issues)
Increase by 7% in 5 years
Edmonton City is the capital city of the Canadian province of Alberta located in the middle west of the country. The city is relatively wealthy in economy, and there has been many migrants from neighboring provinces and immigrants, inflating city’s population by 7.2% during recent 7 years to 960,000. On the other hand they believe new industries and business are needed to support future growth of the city, and, they expect AI related industries and business are suitable for the growth. The major player in “AI Hub” of Edmonton is Amii (Alberta Machine Intelligence Institute).
Amii is famous for its group of 15 globally popular AI researchers of the University of Alberta represented by Professor Richards Sutton considered one of the founding fathers of modern computational reinforcement learning. Amii was recognized and selected as the first overseas institute by British artificial intelligence company DeepMind which is famous for its AlphaGO GO AI program beating GO world-champion.
Amii is advanced in other field than gaming, and PH (Precision Health) can be named as one example utilizing AI technology. Precision health is next-generation preventive medicine, taking big data but highly personalized approach to predict disease occurrence of each individual patient and prevent and treat the disease at earlier stage.
Since 3 years Amii has been studying in cooperation with school of Medicine and Dentistry of the University of Alberta to achieve earlier detection of the disease based on the predicting modelling of Inflammatory bowel disease and depression. Professor Deborah Jams of the University is expecting that this method of detection could be the main stream of medical prediction and care treatment though practical use is a little away in the future.
Accumulation of wide variety of Knowledge
Amii has been accumulating and storing a wide variety of knowledge concerning many cases of utilization of AI from academic to practical aspects through its deep involvement in industrial applications. Professor Randy Goebel points out that continuous learning system is indispensable to enhance prediction model when AI is employed in the development of prediction models not limited to PH (Precision Health).
Noteworthy point is that in his idea a series of flows are repeatedly executed after clear reasoning is given to the validity of the action result based on the prediction model. This flow and process is absolutely necessary in the actual adoption and application of AI in the real society.
Theoretical Insight
The professor is further giving advice that following three facts are important in establishing proper learning system.
① “AI expert” needs theoretical insight.
When asked about the volume of necessary data, the professor explained the case that accuracy level of Image Diagnostic model of MRI (Magnetic Resonance Imaging) which learned from only 55 actual data records up to 96%. He further commented “If anyone claims that millions of examples of labeled data are required, I would allege that they have no theoretical insight, and are reporting only empirical results.”
②“Data Access” is the biggest issue commonly shared internationally.
The first issue is the proper consideration to the individual’s privacy. Next comes the issue whether access can be secured to the right data at right appropriate time. In the area of PH this is the very critical issue affecting the robust of the model.
③ Experts in each domain: How the model can be designed and completed as desired is fully dependent on how the knowledge of experts in each domain can be incorporated. The professor stressed the importance that the trusted relationship is established with AI experts and the mutual appreciation and respect is given at least to the study field or domain of each expert.
As reported above, Amii is the strong advantage for Edmonton City to promote AI Industrialization.
