Page 158 - ITU KALEIDOSCOPE, ATLANTA 2019

2019 ITU Kaleidoscope Academic Conference

4. KEY TRENDS IN AI ERA

Following the development track, AI is seen as one of the
most prominent technologies in the intelligent stage, as is
also reflected by the national strategy documents of countries.
The following sections will extract the interaction of AI on
digital health separately from the intelligence stage and
discuss the key trends from the perspective of the main
component factors of AI. They are data, computing power
and algorithms which can correspond to the data, platform
and application of the ICT part in the previous framework.

4.1 Comprehensive description of health data
Figure 4 – Interaction between ICT and health at Stage 2
Large amounts of data are the foundation of intelligent
3.3 Service intelligentization services. In order to more fully describe the state of human
health, two dimensions of expansion are undertaken,
horizontally and vertically.
The third stage is service intelligentization. In the previous
stage, the digitization of health records laid a good
foundation of data sharing and intelligent services for a wider Horizontal expansion refers to the full coverage of a life
cycle. With the keen perception of sensors and strong
range [23]. Figure 5 shows the interaction in blue. Data,
computing platforms and personalized applications are the analytical ability of AI, it could ideally cover the whole
process of user life, continuously monitoring and
main factors to promote service intelligentization. Data is not
limited to the digitization of records, but also refers to comprehensively analyzing various data indicators,
including physiological data (such as blood pressure, pulse),
emerging big data technology, such as IBM Watson built on
big data analysis. Computing platforms are to support the environmental data (such as air that is breathed in), behavior
data (such as exercising or diet), etc. IBM Watson and
process of ‘massive’ EHR and mining the hidden values.
With the increasing volume and complexity of patient Microsoft Azure have built a population health platform
based on “AI+Cloud”, providing an overview analysis of
information, the expectations for rapid and accurate
diagnosis and treatment also rises. AI/ML has great potential various impact factors on personal health. Potential
stakeholders including wearables companies, medical
to assist physicians with reference diagnosis and
personalized treatment. An evidence-based medical institutions, HIS developers and health insurance, etc. can all
benefit from this model. From “treat diseases” to “prevent
decision-making system was established with the help of a
large number of cancer clinical knowledge, molecular and diseases”, it will to some extent alleviate the gap between
supply and demand, mentioned in section 1.
genomic data and cancer case history information [24].
DeepMind also stepped into the AI for health field and
announced its first major health project in 2016: a Vertical expansion refers to the deep description of life.
collaboration with the Royal Free London NHS Foundation Measurement technology is continuously evolving, from the
Trust, to assist in the management of acute kidney injury [25]. individual level, anatomical level, human tissue, metabolism,
Not only diagnosis and treatment are penetrated, intelligent to protein, genetic aspects. Precision medicine was proposed
applications can be integrated in every part of the service with the rapid advancement of genome sequencing
chain, and the corresponding application comes into being, technology and the cross-application of big data technology.
which will be illustrated in section 4.3. The United States initially invested $215 million in the
Precision Medicine Initiative, China has planned to invest
US$9 billion and mentioned precision medicine in the “13th
Five-Year Plan”; Australia launched the Zero Childhood
Cancer Program in 2016 with an investment of A$20 million;
the French genome medical treatment 2025 was also
launched with an investment of 670 million euros. As the
granularity of health data descriptions deepens, AI is able to
establish an interpretation bridge between genetic
information and clinical characterization, and ultimately
achieve personalized and precise treatment.

4.2 Customized computing abilities for scenarios

With rapid increases in the amount and complexity of health
data, higher requirements are proposed for the platform. Two
Figure 5 – Interaction between ICT and health at Stage 3 ways for improvement are: processors and architecture.

– 138 –