Artificial Intelligence in Healthcare and Education
Research Guide

Research Sub-Topics

Why It Matters

In healthcare, AI is used for clinically consequential tasks such as diagnosis, prediction, and decision support, but these uses raise accountability and safety questions that make explainability and interpretability central concerns. Davenport and Kalakota (2019) classify healthcare AI applications into categories that “involve diagnosis and treatment recommendations,” situating AI as a tool used by payers, providers, and life-sciences companies rather than only as a research prototype. Topol (2018) argues in "High-performance medicine: the convergence of human and artificial intelligence" (2018) that the practical goal is a convergence of clinician expertise and AI capabilities, implying that clinical workflows and training must adapt together. In education, large language models are framed as both opportunity and risk: Kasneci et al. (2023) in "ChatGPT for good? On opportunities and challenges of large language models for education" (2023) analyzes how systems like ChatGPT can support learning while introducing challenges that educators and institutions must manage. A concrete, high-stakes example of why this matters is the requirement for responsible decision-making when AI influences patient care or student progression: Rudin (2019) in "Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead" (2019) argues that, for high-stakes decisions, practitioners should prefer interpretable models over post-hoc explanations of black boxes, directly affecting how clinical decision support tools are selected and how assessment-related educational AI is governed.

Reading Guide

Papers at a Glance

In the News

Code & Tools

Recent Preprints

Latest Developments

Frequently Asked Questions

Open Research Questions

• ? Which XAI approaches described in "Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI" (2019) can be operationalized as acceptance criteria for clinical decision support tools without relying on post-hoc explanations of black-box models criticized in "Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead" (2019)?
• ? How can healthcare deep-learning development practices discussed in "A guide to deep learning in healthcare" (2018) be aligned with the human–AI workflow convergence goals articulated in "High-performance medicine: the convergence of human and artificial intelligence" (2018) to reduce deployment failure in real clinical settings?
• ? Which evaluation designs can jointly measure educational benefit and risk for large language models, consistent with the opportunities-and-challenges framing in "ChatGPT for good? On opportunities and challenges of large language models for education" (2023) and the educator-participation concerns raised in "Systematic review of research on artificial intelligence applications in higher education – where are the educators?" (2019)?
• ? What governance and accountability mechanisms are needed when AI systems influence both patient care and educational assessment, given the black-box impediments surveyed in "Peeking Inside the Black-Box: A Survey on Explainable Artificial Intelligence (XAI)" (2018) and the high-stakes interpretability argument in "Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead" (2019)?

Research Artificial Intelligence in Healthcare and Education with AI

PapersFlow provides specialized AI tools for your field researchers. Here are the most relevant for this topic: