Page 35 - Shaping ethics, regulation and standardization in AI for health

Shaping ethics, regulation and standardization in AI for health

symptom assessment applications that have become widely available. These systems, also called
"symptom-checkers", allow their users to enter presenting complaints they seek advice for. The
systems then follow-up with a conversation collecting further evidence on other symptoms
the user might have experienced to then provide advice on relevant next steps ranging from
self-care, oversee a pharmacy to seek emergency care, diseases that might have caused the
symptoms and explanations on how the symptoms and these suggestions are related. By
navigating users to the right care at the right time such systems help using the resources of the
health systems more efficient. On the doctor's side such systems help to save time by allowing
for an automated collection of relevant information before seeing the doctor and to reduce the
risk of misdiagnosis. While systems for AI-based symptom assessment have great potential to
improve health care, the lack of consistent standardisation makes it difficult for organizations
like the WHO, governments and other key players to adopt such applications as part of their
policies to address global health challenges. The specification of a standardized benchmarking
for AI based symptom assessment applications in this document as part of the ITU/WHO Focus
Group on AI for Health is an important step towards closing this gap.

A�4�12 DEL 10�15: FG-AI4H Topic Description Document for the Topic group
on Tuberculosis (TG-TB)

Summary: This TDD focuses on the use of AI for the radiographic detection and screening
of tuberculosis (TB), particularly in high-burden areas like India. It highlights the potential of
AI, specifically Computer Assisted Diagnosis (CAD) systems, to overcome healthcare worker
shortages and improve early TB detection, which is crucial for global TB control efforts.
The core purpose of this document is to propose and detail a standardized benchmarking
approach for evaluating the performance of AI-based TB screening tools, including defining
data requirements, performance metrics, and the collaborative process for developing and
testing these tools.

A�4�13 DEL 10�17: FG-AI4H Topic Description Document for the Topic group
on dental diagnostics and digital dentistry (TG-Dental)

Summary: This TDD details the work of the Topic Group on Dental Diagnostics and Digital
Dentistry (TG-Dental). It provides a comprehensive overview of the group's activities from
2019 to 2023, outlining the challenges and opportunities of AI in various dental specialties,
including diagnostics, treatment planning, and digital dentistry. The text emphasizes the need
for standardized benchmarking of AI systems to ensure their robustness and generalizability
across diverse populations and clinical settings, highlighting the ethical considerations, such as
data diversity and privacy, that are crucial for the responsible development and implementation
of dental AI solutions.

A�4�14 DEL 10�20: FG-AI4H Topic Description Document for the Topic group
on AI for endoscopy (TG-Endoscopy)

Summary: Endoscopy is the core technical means for early diagnosis and screening of digestive
cancer, while AI solutions for endoscopy are expected to help clinicians improve the quality
of their examinations and reduce the number of missed diagnoses. This TDD describes the
application of AI in endoscopic procedures, specifically focusing on two subtopics: colonoscopy
and endoscopic ultrasound (EUS). In addition to a general description of AI for endoscopy,
this document defines a framework for standardized benchmarking of AI systems designed to

25