Datasets:
lfischbe
/
diss

Dataset card Data Studio Files Community
2
text
stringlengths
1
1.75k
Aporia
A performative statement of doubt, uncertainty, or perplexity.
Artificial Intelligence (AI)
Artificial Intelligence (AI) is a widely used term to describe any tool or system that simulates, augments, or automates the way people make sense of the world. Legally, AI has been defined in the US as
“A machine-based system that can, for a given set of human-defined objectives, make predictions, recommendations or decisions influencing real or virtual environments. Artificial intelligence systems use machine and human-based inputs to- (A) perceive real and virtual environments; (B) abstract such perceptions into models through analysis in an automated manner; and (C) use model inference to formulate options for information or action.”
AI encompasses deep learning as well as machine learning, and some, but not all, natural language processing and robotics (embodied AI). In the present historical moment a secondary definition is ascendant, in which AI refers broadly to any technology for predictive pattern matching. In its contemporary use of the term, AI is more likely to refer to generative, unsupervised, generalizable, and stochastic processes of analysis and resynthesis that can be applied across modalities, rather than structured simulations optimized for singular, specific tasks. (See: Apperception, Algorithmic Violence, Bias, Diffusion, Embedding, Latent Space, Machine Learning, Inference, Large Language Model (LLM), Model, Variational Autoencoder (VAE))
Attention
Focusing on some information while ignoring the rest. Attention is a fundamental cognitive process involved in perception, learning, memory, and decision-making. (See: Attunement, Noise, Signal, Transformer, Waste)
Attunement
Affective alignment between individuals. Attunement is a relational foundation for trust and understanding, characterized by attention to cues (verbal and non-verbal, body language, tone, expression, etc) and the ability to perceive and understand one’s own emotions and those of others. (See: Affect, Attention, Coherent, Consent, Diffusion, Listening, Resonance, Soundscape)
Autoencoder
(See: Variational Autoencoder (VAE))
Automated Decision-Making (ADM)
Decisions made on the basis of automated processes, including the derivation of profiles based on perceived demographic or behavioral patterns related to a subject. ADM is restricted under various legal frameworks: European Union law preserves the rights of individuals to seek human intervention in decisions, while US law offers specific prohibitions on the use of ADM such as access to housing or public assistance, and outlines a path for ongoing audit of automated decision-making process for bias and lack of transparency. (See: Artificial Intelligence (AI), Algorithmic Violence, Predictive Decision Support Intervention (DSI))
Autopoiesis
Making, reproducing, and sustaining one’s self. An autopoetic organism, self, machine or system consists of a “network of processes” which, “through their interactions and transformations regenerate and realize the network of processes (relations) that produced them.”
Biographic Mediation
The instrumentalization of life-writing in seeking aid: how we tell our stories in order to receive care. (See: Anamnesis, Instrument, Hypomnemata, Life-Writing)
Biopsychosocial Model
First proposed in the mid-1970’s and widely adopted by clinicians, researchers, and educators over the ensuing generation, the biopsychosocial model situates the complex interface between medical knowledge and the needs of individual patients in “a way of understanding how suffering, disease, and illness are affected by multiple levels of organization, from the societal to the molecular,” while simultaneously centering “the patient’s subjective experience as an essential contributor to accurate diagnosis, health outcomes, and humane care.” Put simply, “clinicians must attend simultaneously to the biological, psychological, and social dimensions of illness” in order “to understand and respond adequately to patients’ suffering—and to give them a sense of being understood.” Linked to systems theory in its emphasis of context and cross-disciplinarity over analytic approaches characterized by narrow specialization, the biopsychosocial model seeks to preserve complex, holistic relationships in its analyses. (See: Bodymind, Exposome)
Bodymind
A vector in disability discourse away from dualist (e.g. Cartesian) separations of bodily and mental knowledge, experience, causes and effects: “We can refer meaningfully, if tentatively, to "mind" and "body," but ultimately the two are so fully integrated that they should also be considered one,” in that “mental and physical processes not only affect each other but also give rise to each other.” (See: Biopsychosocial Model)
Coherent
Like two waves arriving at the shoreline in the same moment. (See: Signal)
Common Sense
The way our own senses fit together with one another, and become legible to others. As Arendt uses the term, it is protection from alienation, both politically and bodily: “It is by virtue of common sense that the other sense perceptions are known to disclose reality and are not merely felt as irritations of our nerves or resistance sensations of our bodies.” (See: Discourse, Disidentification, Epistemology, Ideology, Interpellation)
Composition
An action of “re-listening:” a set of collectively defined procedures that are iteratively applied, tested, and reworked. Making a form, not like building a house out of building materials (construction), but like describing something to create a mental picture, or getting in the habit of exercising. (See: Listening, Poetics, Soundscape)
Compression
In terms of representation, compression is a process of making a simpler or more lightweight representation from an original version. A lossy compression is created, in part, by removing information that is not deemed meaningful, such that the reconstituted copy has lost some aspect of the original. Compression is a fundamental process in digital representations, including AI. Conceptually, compression can serve as a metaphor for any simplified communication of complex reality: thought into language, an explanation, a diagram, etc. (See: Embedding, Information, Latent Space, Sampling, Variational Autoencoder)
Consent
A negotiation of agency, concerning both what others allow us to do to them, and what we allow others to do to us.
Crip
Crip articulates pride, critical awareness, and celebration of difference with regard to identifying as a disabled person. Similar to queer in that it is a slur reclaimed as a term of self-identity, crip expresses the non-compliant, anti-assimilationist position that disability is a desirable part of the world. When used as a verb, to crip is to engage in “practices of critique, alteration, and reinvention of our material-discursive world,” to re-center narratives of disability in contexts where it has been overlooked or excluded. (See: Accessibility, Disability, Disability Justice, Impairment)
Data
Raw stuff that, through a process of interpretation, becomes information that may have meaning. Can have its own structure and context, or not. Can be material or abstract. Can be of one kind or many. (See: Ground Truth, Information, Interpretation)
Debility
A state of weakness, incapacity, or impairment, limiting an individual's ability to perform activities of daily life. Jasbir Puar situates debility as a spectrum of experiences that include the dynamics of vulnerability and resilience, exploring how certain bodies are produced as debilitated through systems of oppression, globalization, and the politics of health. (See: Disability, Impairment, Normal)
Design
“The interaction of understanding and creation,” including but not limited to methodologies where design is a conscious strategy, also implicit in “how a society engenders inventions whose existence in turn alters that society.” (See: Sociotechnical)
Diagnosis (Dx)
The action of identifying patterns in an individual’s health status so as to assign categorical labels within agreed-on institutional systems. Diagnosis may take into account the individual’s health history, test results and other measurements. (See: Data, History (Hx), Information, Treatment (Tx))
Diffusion
Diffusion models, used in generative AI to produce realistic synthetic images and data, destroy the structure of the original (what is recognizable about it) by adding noise, then learn how to iteratively restore the original from the noise. (See: Compression, Composition, Coherent, Distortion, Noise, Variational Autoencoder (VAE))
Digital Biomarkers
As defined by the eponymous academic journal, Digital Biomarkers are “objective, quantifiable physiological and behavioral data that are collected and measured by means of digital devices such as portables, wearables, implantables, or ingestibles. The data collected are typically used to explain, influence, and/or predict health-related outcomes.”
Digital Phenotype
Patterns in data, correlated with health-related outcomes. (See: Data, Diagnosis, Digital Biomarkers, Outcome)
Disability
Disability is an institutional category with social, legal and political consequences. It has been theorized in medical (in terms of pathology and cure), social (in terms of constructed exclusions and social change), as well as political/relational terms (medicine is political, social status is intersectional, embodied knowledge is valuable). Disability can also be a part of a person’s identity. Allison Kafer writes: “ideas about disability [...] animate many of our collective evocations of the future; in these imaginings, disability too often serves as the agreed-upon limit of our projected futures.” As a function of social imagination, it informs what we think is possible. Disability provides a legal and ethical framework for identifying aspects of the environment that exclude or impede people with impairments, and to improve the contexts—legal, social, and architectural, etc—in which disabled bodies and minds exist. While these frameworks overlap conceptually and practically with those of illness—in that some illnesses may produce disability, and some disabilities may produce illness—the two terms are not interchangeable. (See: Ableism, Accessibility, Bodymind, Crip, Disability Justice, Illness, Imaginary, Impairment, Protected Attributes)
Disability Justice (DJ)
An organizing framework for exposing and challenging ableism and other forms of oppression that impact people with disabilities. Building on critical and legal concepts of Disability Rights (self-determination, autonomy, inclusion, access to resources, support, and opportunities to thrive), DJ places renewed focus on how disability intersects with race, gender, sexuality, class, and other social identities. (See: Ableism, Accessibility, Crip, Disability, Impairment)
Discourse
Where thought is expressed in language, shaping a shared sense of what is appropriate, normal, true, or possible. (See: Common Sense, Disidentification, Epistemology, Ideology, Imaginary, Interpellation)
Disidentification
The productively ambiguous position of neither assimilating to, nor reacting against, dominant patterns. (See: Common Sense, Discourse, Ideology, Imaginary, Interpellation)
Distortion
What happens when a thing is deformed past a threshold of recognizability or translation. In terms of signals, distortion is the process of progressively introducing harmonics, adding complexity until the signal becomes indistinguishable from noise.
Easy Read
A way of writing that helps people to understand the main ideas more clearly. Easy read uses short sentences and words, and also has a picture next to each sentence. The picture helps give more information about the words. (See: Accessibility, Information, Plain Language)
Ecological
Having to do with the study of relationships, between humans, cultures, systems, processes, tools, and/or non-humans. (See: Clinic, Home, Environmental, In The Wild)
Electronic Health Records (EHR)
including laboratory and test results, lists of prescribed medications, vaccinations, and narrative reports written by clinicians (commonly referred to as “open notes”). These records may be accessible to patients without mediation or interpretation
Embedding
A way of representing complex data in a simpler form, while preserving important properties and relationships. An image of a face is reduced to a collection of nodes and edges. No dictionary definitions are explicitly learned, but a word is understood in the number of times it appears in the vicinity of other words. A basic example is the task of representing a city—a highly complex entity—in terms of its latitude and longitude, two vectors that identify the city by where on the globe it is located. This may be all you need to know about the city. Adding more details—population, country, yearly precipitation, median income, and so on—builds a richer representation, while adding more dimensions to the embedding space. AI models given the task of creating embeddings learn which implicit patterns and relationships in the data are important to know, without need for high fidelity representations of the individual items. (See: Artificial Intelligence, Compression, Large Language Model (LLM), Machine Learning, Variational Autoencoder (VAE), Retrieval Augmented Generation (RAG))
Embodiment
Being in one’s skin. “the lived body as, at once, both an objective subject and a subjective object: a sentient, sensual, and sensible ensemble of materialized capacities and agency that literally and figurally makes sense of, and to, both ourselves and others.” What Donna Haraway called, non-metaphorically, “significant prosthesis:” a vector of meaningfulness that connects between flesh and environment. Occupying a coherent position, orientation, or accountability, and as such, having agential status in political, legal, ontological and epistemological terms. (See: Coherent, Environment, Epistemology)
Emplotment
Organizing a series of events into a narrative with a plot, giving it structure, coherence, and causality. Used to contextualize events into meaningful totalities, and to give meaning to disparate sets of events. (See: Ideology, Hermeneutics, Explainability)
Endpoint
In research, as in clinical care, an endpoint is identified as significant and relevant information produced by processing raw data, or outcomes. (See: Data, Information, Outcome)
Environmental
In healthcare, environment refers to the interaction of external factors on a person’s well-being and health outcomes. Environment may include physical, biological, social, cultural, behavioral, political and regulatory factors. Environment may contain multiple ecologies. (See: Clinic, Home, Ecological, Exposome, In The Wild)
Epistemology
How we know what is true, how we know anything at all. Ideas, changing and evolving, about how knowing works, where it comes from, what it covers, and what it can and can’t do. (See: Discourse, Ground Truth)
Explainability
In the context of artificial intelligence, a system can be said to be explainable if it includes methods to assist humans who interact with it to understand what automated steps were taken to reach an outcome, even if the low-level mechanics of the solution are inaccessible, as with deep learning models. Explainability often takes shape as a summary or high level overview of the processes and justifications used. (See: Artificial Intelligence, Deep Learning, Emplotment, Interpretability, Large Language Model (LLM), Machine Learning, Model)
Exposome
A “comprehensive description of lifelong exposure history” that complements the genome, encompassing internal processes (from gut flora to hormones), as well as both specific and general external factors (pollution, climate, education, class, stress, and so on). (See: Anamnesis, Embodiment, Environmental, History (Hx), Life Writing)
Fairness
In the context of AI and machine learning, fairness means freedom from error when measuring between protected attributes—the set of personal characteristics that are protected from discrimination by law and cannot be used as the basis for decisions, such as race, gender, disability status, and so on. (See: Accuracy, Automated Decision-Making (ADM))
Generative AI
Simply, a kind of AI model that can make new data based on patterns and relationships learned in its training data. (See: Artificial Intelligence (AI), Variational Autoencoder (VAE)
Ground Truth
What is agreed to be stable and true, before any analysis or modeling takes place. The actual, confirmed reality or facts regarding a specific situation, used as a comparison or benchmark to show that models, analyses, and derived data are valid and accurate. (See: Data, Discourse, Ideology, Model)
Health-related Quality of Life (HRQL)
An important measure of patient-centered outcomes. The key domains of HRQoL include behavior (what you do), functional status (what you can do), symptoms and symptom burden (how you feel and how it impacts what you can do).
Hermeneutics
The study of interpretation. WIth applications from literary theory to ethnography, a set of methods that prepare an observer from one tradition to understand or describe—but not try to explain—the practices and languages of another, including gathering awareness of what is taken for granted within the observers’ own tradition. (See: Incommensurability, Interpretability)
History (Hx)
In a healthcare context, History, abbreviated as Hx, is a record of information about a person’s health both current and in the past. A person’s Hx records may include information about their allergies, illnesses, surgeries, immunizations, results of physical exams and tests, medicines taken, as well as habits and behaviors. (See: Anamnesis)
Home
In the design of research studies (e.g. clinical trials), home refers to experimental settings that are intended to capture more accurate data about subjects’ daily life than is possible in a clinic. A number of assumptions may be embedded in the notion of home, such as assuming that subjects have access to a stable, consistent and safe environment where they have the ability to engage with the research in a self-supervised way. (See: Clinic, Ecological, Environmental, In The Wild)
Homotopy
In the context of generative AI, different samples taken from the same model’s latent space share a homotopic relationship, in that each can be continuously deformed into the other along the contours of the feature space. For example: a pair of homotopic images of cats may be sampled from the latent space of a model trained on images of cats. By interpolating between the two images, a series of new images can be continuously sampled, where the first image of a cat appears to change gradually into the second, without ever presenting as either non-image or non-cat. It is likely that the latent space illuminated by this model does not hold all possible cats, but everything it holds is likely to look like a cat. This is evidence that the model has learned a stable, persistent representation of the visual structure of cats. (See: Interpolation, Latent Space)
Hypomnemata
(also written as Hupomnemata) Ὑπομνήματα [notebooks]. Collected notes and reflections “which must be reread from time to time so as to reactualize their contents,” alongside and intertwined with practices of listening and self-reflection. From Michel Foucault’s study of Seneca’s rules for self-knowledge and self-care. (See: Imaginary, Life-Writing)
Ideology
Dominant patterns that apply coherence, as the influence of power through groups and systems: “The imaginary relationship of individuals to their real conditions of existence.” (See: Common Sense, Discourse, Disidentification, Imaginary, Interpellation)