Skip to main content
Trustworthy AI for Dementia
Mah Parsa
Presenting author:
Mah Parsa
Growing user-friendly machine learning and deep learning libraries and powerful computational resources, graphics processing units (GPU) and tensor processing units (TPU) have motivated AI developers to develop AI-powered systems to help clinicians to detect dementia and diagnose subtypes of dementia or identify older adults at risk of dementia.
Developing such AI systems can be beneficial for different levels of societies, including individuals and healthcare communities, but it can also damage human rights and healthcare communities' reputations. In other words, developing AI for dementia is beneficial for the elderly by offering them "the right to know" and providing them quick access to disease detection settings and chances to plan their futures. But using AI to detect dementia could be harmful to older adults due to the lack of disease treatments and biased results (i.e., any misdiagnosed outcomes of AI systems can harm the social rights of older adults). Thus developing trustworthy AI for dementia is essential. Developing such systems bring opportunities for the medical community but raises challenges for the AI community. The main challenge is the lack of a unified framework to establish a trustworthy AI for dementia.

General speaking, a trustworthy AI system has been defined as an AI system that not only respects all applicable laws and regulations but also should:
Support fundamental human rights.
Be robust, safe, transparent, and fair.
Not decrease, limit or misguide human autonomy.
Follow diversity principles.
Enhance positive social change.

Thus, to develop trustworthy AI for dementia, AI developers should follow the above criteria and consider the perspectives of clinicians and mental health professionals, caregivers, patients, and their family members, other AI developers, and social systems about their developed AI systems.
Even though the main benefit of developing trustworthy AI-powered systems for dementia is to encourage clinicians to adapt them in clinical settings for detecting dementia, to convince patients to accept the outcomes of AI-powered detection systems for dementia, AI developers should also consider ethical principles.