AI-enhanced prognostication in cardiac resynchronization therapy using displacement encoding with stimulated echoes (DENSE) MRI;Reimagining Patient Care: The Socio-Technical Dimensions of AI in Healthcare

Artificial intelligence (AI) has revolutionized healthcare, offering breakthroughs in medical decision-making, but it also raises serious questions about balancing human judgment with algorithmic recommendations. My capstone project explores this duality by developing a predictive model for Cardiac Resynchronization Therapy (CRT), while my STS research critically examines how AI influences physician authority and patient trust. Together, these projects investigate both the technical and ethical challenges of integrating AI into patient care. While the capstone aims to optimize outcomes using advanced imaging and deep learning, the STS component evaluates whether these innovations enhance or erode the human aspects of medicine. The capstone project centers on creating a 3D convolutional autoencoder that uses Displacement Encoding with Stimulated Echoes (DENSE) MRI data to predict CRT response. Since nearly one-third of CRT patients currently fail to benefit from the therapy, our aim was to reduce this non-responder rate through better patient selection. The model is trained on myocardial displacement fields and paired with a classification output that estimates therapy success. By integrating rich imaging data and deep learning, this approach supports more personalized, data-driven treatment decisions for heart failure patients. Initial results demonstrated strong classification accuracy, with the model reaching validation accuracy scores near 0.70, outperforming traditional ECG-based criteria. These findings suggest that AI can substantially improve CRT success rates by enabling more targeted therapy. However, limitations remain. Reconstruction quality was inconsistent, and interpretability issues highlight the need for transparent models. The project concludes that AI, when combined with DENSE MRI, can enhance cardiac care, provided that clinicians remain actively engaged in the decision-making process and ethical concerns are addressed. My STS research complements the capstone by examining the question: How does AI affect clinician autonomy and patient trust in high-stakes medical decisions? Using the Ethics of Care framework, which emphasizes empathy, relational responsibility, and contextual sensitivity, I assessed how AI tools might alter the interpersonal dynamics of healthcare. While AI offers efficiency and precision, its opaque “black box” nature risks distancing clinicians from their patients and patients from their care. By grounding the analysis in real-world clinical scenarios, I explored how AI shifts the authority structure in medicine and impacts the emotional and communicative dimensions of care. Through scholarly analysis and healthcare case studies, I found that ethical integration of AI depends on transparency, shared decision-making, and preserving human discretion. Clinicians are often wary of systems they cannot interpret, and patients are less likely to trust AI recommendations when they are not clearly explained. The Ethics of Care urges that AI be used as a collaborative tool, not an infallible replacement, within a system designed to prioritize dialogue and compassion. If implemented carefully, AI can support more just and effective care. But without clear safeguards, it risks displacing the very relationships that define good medicine. Together, my technical and STS projects offer a vision for AI in healthcare that is both innovative and deeply human-centered. Together, my capstone project and STS research form a coherent investigation into how AI can improve patient outcomes while safeguarding human-centered care. By combining technical innovation with ethical scrutiny, I hope to contribute to a healthcare system that is both technologically advanced and deeply compassionate.