#Digital_Phenotype Read this on [[10-07-2024]] Related to - [[Book Notes - Human Compatible- AI and the Problem of Control]] - #humanism - [[Digital Phenotyping]] - [[Digital Interventions]] - [[Design]] - [[Principles in designing Digital interventions]] - [[First-person disavowals of digital phenotyping and epistemic injustice in psychiatry]] - [[Paper - From Digital Phenotype Identification To Detection Of Psychotic Relapses]] - [[With Power Comes Responsibility]] ![[image 1.png]] ![[Aesthetics of algorithmic care.pdf]] My Notes 1. I am curious - what are the "data ethics considerations", "Philosophy", "Social science" views or challenges to the epistemology and methodology of digital phenotyping? 2. Machine learning, requires large amount of data that are "cleaned", "standardised", but how does that take into consideration of the unique human experiences that are culturally unique? 1. Health expressed differently according to culture/social? What is health in one culture is it different in another? How is illness talked about, expressed, detected are different? 3. Reminded me of =="To have a humanist’s heart, a scientist’s head, and an artist’s eyes” - Liu Thai Ker"== -- so call human-centric design 1. "Design as understood here, is necessary to understand the needs of users and relations within user's ecologies and thus define affordances, functions and interactions" 2. Here, aesthetics involves creating systems that are not only visually appealing but also thoughtfully designed to foster positive, ethical, and compassionate experiences. This includes considering how the system presents information, engages with users, and supports their mental health in a respectful and empathetic way. Pragmatist aesthetics and aesthetics of care emphasize practical and caring approaches to design, ensuring the technology is user-centered and promotes well-being. 4. We cannot reduce human to signals. Behavioural is not just about symptoms, or many reasons for behaviour 1. “Digital phenotypes can oversimplify people by interpreting all behaviors as potential symptoms using AI. This means that digital phenotyping can shape how people see themselves, their bodies, and minds, and influence their thoughts and actions. Users might also change their behavior, consciously or unconsciously, to influence the data. Therefore, developers of digital phenotyping systems must be careful about how they handle, explain, and present mental health information to users.” 5. How to balance between the need to standardise data and being sensitive to context. 6. We must not think we know everything, and decide that our protocol or our rules are the truth. telling people how to behavior or do. 7. The Four Domains to considering the design 1. Perception (Including assumptions of the codes) 2. Representation (UI UX) 3. Experience (Mental Health Models) -- 4. Ecologies --- How is this service/design embed in a larger picture/system?