Farjana's Technical Blog

Designing Epistemic Virtues into Privacy Preserving Systems

Building on previous posts about trust erosion in privacy-preserving health data, this piece argues for designing systems that cultivate epistemic virtues intellectual humility, courage, and responsibility rather than treating privacy as a simple trade-off. A practical four-step framework shows how to embed these virtues from the start, ensuring systems built for places like Bangladesh protect both data and the quality of knowledge produced.

The Clinician's Black Box: Designing AI That Supports, Not Replaces, Human Judgment

Examining the philosophical clash between clinicians' experiential wisdom and AI's opaque reasoning, this post argues against simply explaining black-box models. Instead, it proposes designing inherently arguable systems using hybrid rule-augmented architectures that foster professional dialogue, preserve clinical autonomy, and measure success through "responsibility-preserving performance."

Auditing Trust in Health Registries: From Technical Fairness to Epistemic Responsibility

Building on previous work about ethical auditing, this post argues that real trust in health data systems requires moving beyond technical compliance to embrace epistemic responsibility. Exploring how to audit not just data but knowledge itself, we examine frameworks for multidimensional fairness, transparent uncertainty reporting, and the steward's duty to prevent misleading insights in compliant but flawed systems.

Differential Privacy Noise and Clinical Epistemic Trust: Can We Have Both?

Exploring how the very noise that protects patient privacy in health registries can erode clinicians' ability to form justified beliefs about medical reality. This post examines the epistemic disconnect between privacy mechanisms and clinical knowledge, proposing transparent systems with selective fidelity and clear trust calibration for rare disease monitoring and public health decision-making.

Balancing Privacy and Explainability: A Technical and Epistemic Challenge for Clinical Data Sharing

Examining the fundamental tension between Differential Privacy's mathematical guarantees and the need for transparent explanations in clinical AI. This post explores how privacy-preserving noise conflicts with explainable AI methods, proposing a co-design framework that integrates hybrid architectures, selective DP application, and social-epistemological principles for building trustworthy health data ecosystems.

From Technical Privacy to Ethical Accountability: Auditing Federated Health AI

Exploring why federated learning's privacy guarantees are insufficient without systematic ethical audits. Drawing from ECG analysis and registry projects, this post presents a practical framework combining rule-augmented networks, multiparty differential privacy, and continuous fairness monitoring to build truly accountable health AI systems.

Differential Privacy in Health Registries: Protecting Data Without Losing Clinical Insight

Exploring the ethical and technical challenges of applying differential privacy to sensitive health data. How traditional approaches can erase rare disease signals and marginalized communities, and the adaptive methods that protect privacy while preserving clinical truth and equity.

Beyond Scaling: What Ilya Sutskever's AI Warning Means for Global Health

A personal reflection on Ilya Sutskever's podcast insights about AI's limitations and the transition from scaling to true understanding. Connecting his observations about "shallow understanding" and learning efficiency to real challenges in global health AI, from dengue symptom triage in Bangladesh to fairness-aware ECG analysis.

The Future of Trustworthy AI: Why Hybrid Intelligence is the Key to Human Confidence

Exploring how hybrid symbolic-neural models are solving AI's black box problem by combining neural network pattern recognition with symbolic AI's logical reasoning. Drawing from my research in healthcare AI and fairness-aware systems, this article examines the technical architecture and ethical implications of building AI that doesn't just perform well but explains itself and earns genuine trust.

Personal Reflection: From 0s and 1s to Life's Big Questions

A heartfelt journey from pure coding to philosophically-informed AI development. Reflecting on how building a dengue symptom triage chatbot revealed the ethical dimensions of technical decisions, and how concepts like epistemic injustice transformed my approach to creating trustworthy, human-centered AI systems.

From Philosophical Guardrails to Clinical Black Boxes: Building Ethical AI in an Opaque World

Examining the critical gap between AI ethical training and real-world vulnerabilities through Anthropic's security incidents. This analysis connects state-sponsored AI manipulation to epistemic opacity challenges in clinical decision-support, proposing a framework for healthcare AI that is both powerful and trustworthy. Includes references to 60 Minutes interviews and security disclosures.

The Cultural Calculus of Care: How Economics and Values Shape Global Healthcare AI

Exploring how cultural contexts and economic realities create divergent AI adoption paths in healthcare from Scandinavia's privacy-focused precision medicine to Bangladesh's access-driven triage systems. Drawing from my research on dengue symptom triage AI and fairness-aware ECG analysis, this article examines the ethical tightrope between data privacy and equitable access in global health AI deployment.

AI Explainability Tools: A Hands-On Review of SHAP & LIME

A practical deep dive into SHAP and LIME explainability tools based on real research in intersectional fairness. This review includes hands-on code examples, critical analysis of limitations, and insights about moving beyond technical explanations to build truly accountable AI systems. Learn how these tools revealed critical fairness violations in image classification models.

Ethical AI in Practice: Diagnosing and Curing Algorithmic Bias in Healthcare

When AI models are trained on data reflecting historical human biases, they don't just learn medicine they learn our inequalities. This analysis examines two critical case studies of algorithmic bias in healthcare AI and proposes an interdisciplinary framework combining technical fairness methods with ethical principles to build more equitable clinical systems.

Hei! My Journey to Learning Norwegian DIAMOND ACHIEVED!

As I prepare for the next chapter of my research career, I decided to start learning Norwegian, one Duolingo streak at a time. Latest (Jan 25): Reached Diamond League with 30000+ XP, 75-day streak (Wildfire EL 7!), Sage achievement earned, and expanded vocabulary for celebrations, directions, and dining!

Healthcare Economics & Sustainability: Lessons from Western Systems

An in-depth analysis of Western healthcare economies through recent research on Norwegian systems. This comprehensive review examines the paradox of high spending with mixed outcomes, explores circular economy implementation in hospital procurement, analyzes strategic hospital design for cost efficiency, and discusses the ethical integration of AI offering transferable principles for global healthcare sustainability.

Trust in AI: Lessons from Healthcare

When a patient is on the operating table, trust is absolute. We trust the surgeon's skill, the anesthesiologist's precision, and the nurse's vigilance. Now, a new team member is entering the room: Artificial Intelligence. From algorithms that detect tumors invisible to the human eye to systems predicting sepsis hours before it strikes, AI's potential to save lives is revolutionary. But in this high-stakes domain, potential is meaningless without a foundation of trust.

Ethical Dilemmas of Bias in AI-Driven Medical Decisions

Imagine two patients with the same symptoms. One is flagged for a life-saving screening; the other is not. The difference isn't their clinical need, but their race or postal code. Now, imagine this decision is made not by a human, but by an algorithm touted for its "objectivity." This is the central ethical crisis of AI in medicine. We risk automating and amplifying the very inequalities the healthcare system has struggled to overcome.

The Hidden Cost of Privacy in Healthcare AI

Exploring the subtle epistemic and social costs of privacy-preserving techniques in healthcare AI. While federated learning and homomorphic encryption protect patient data, they can also change how medical knowledge is produced and shared in clinical settings.

When Accuracy Isn't Enough: What My AI Models Can't Tell a Doctor

High accuracy numbers don't always translate to real-world clinical trust. A reflection on the gap between model metrics and clinician understanding in healthcare AI systems.

Beyond "Explainable AI": What Philosophers Can Teach Us About Trust

Moving beyond technical explanations to philosophical frameworks for building trustworthy AI systems. Exploring epistemic opacity, justification vs. explanation, and designing for different stakeholder needs.

My Recent Research Papers: Advancing AI in Healthcare and Environmental Monitoring

An overview of my latest research contributions in AI-powered healthcare solutions for Bangladesh and environmental monitoring for smart cities. Covering dengue symptom triage chatbots, health indicator analysis, air quality prediction, and fairness in ECG models.

The Secret of the Secrets: Unveiling Hidden Wisdom

A deep dive into Dawn Brown's exploration of hidden wisdom and its connections to cognitive science and AI research.

Welcome to My Technical Blog

I'll be sharing insights from my research in:

• Machine Learning Fairness
• Healthcare AI
• Explainable AI (XAI)
• Federated Learning
• Privacy-Preserving AI
• Ethical AI Development

First Technical Deep Dive

Coming soon: A detailed analysis of gradient sparsification in federated learning with homomorphic encryption.

Stay tuned for more technical content!