Content - Session IV: AI Applications in Health Care

AI Applications in Health Care

Session abstract

Demographic trends provide a huge challenge for health care systems in industrialized countries like Japan, France and Germany. AI has the potential to successfully address some of these challenges like shortage of healthcare professionals, making the rapid growth of medical knowledge available for diagnostic and therapeutic purposes independent of time and location, caring for elderly people in their homes, etc. This session will discuss the most promising fields of application for AI in health and care, how Japan, France and Germany can combine their strengths to compete in the field and in international markets and how the highest quality outcome for patients can be assured.

Artificial Intelligence in Precision Medicine
Chair: Dr. Kazuhiro Sakurada

Precision medicine is a new paradigm that represents a shift from a statistical abstraction of patients toward the view that each patient is unique. This is a new scientific challenge as well as a new social challenge. Biological systems form complex networks where the collective behavior cannot be reduced to simple correlations. To overcome this problem, a new biomedical science based on pure description of diseases is required. For this purpose, we are working to collect and organize, in machine-readable form, “life course data” related to the physical conditions of a person and then leverage the power of artificial intelligence to analyze that data.

Do We Need Quality Standards for AI in Health Care?
Chair: Prof. Dr. Klaus Juffernbruch

AI is a trending topic in medical research with new studies published at a very high rate. The first applications are already approved for clinical use by regulatory bodies like the FDA. Still, it seems that there is research needed on what standards in the evaluation of AI systems could ensure the quality of AI diagnoses and treatment recommendations.

Cognitive Mirroring: Computational Approach to Developmental Disorders
Dr. Yukie Nagai

Developmental disorders such as autism spectrum disorder are characterized by difficulties in social communication. Recent studies, however, suggest that their core problems exist in sensorimotor processing rather than in social cognition. My talk presents our research project titled “Cognitive Mirroring”, which aims at designing artificial systems for investigating the underlying mechanisms of developmental disorders. Our studies using computational neural networks suggest that typical/atypical development can be accounted for by different capabilities in prediction. I will show how artificial systems enable understanding and therefore assist individuals with developmental disorders.

Artificial Intelligence Enables Precision Diagnostics in Clinical Medicine
Prof. Dr. Fabian Kiessling

Significant advances have been achieved in elucidating molecular regulations of diseases and numerous disease-related markers were identified. Additionally, imaging technologies have steadily improved and are providing detailed insight into tissues’ morphology, function and molecular regulation. However, there is still a need to identify and quantify the most relevant information and to bring it into a mechanistic context. Here, I will report on the concept of a comprehensive diagnostic center where medical disciplines closely collaborate with computer scientists to build IT solutions to store, extract, and integrate diagnostic data and to use computer power to build digital disease models.

Using Artificial Intelligence to Personalize Cancer Care
Charlotte Robert & Lucas Fidon

With the advent of retrospective medical data and algorithms for reasoning, understanding and interpreting, precision medicine has become a driving force for a clinical, societal and economic revolution. In this talk I will present the latest results in predicting treatment outcomes in immunotherapy. The field of immunotherapy, which aims to boost the immune system and help it fight cancer, is one of the greatest hopes in modern cancer care. However, it is effective in only 15 to 30 percent of cases. Artificial intelligence makes it possible to predict which patients will benefit from this effective but expensive treatment.