Advancing the Usability of Temporal Stream Logic (TSL)

Temporal Stream Logic (TSL) is a high-level logical specification language that describes the behavior of a reactive system over time. TSL can be synthesized into high-level code such as JavaScript and Python. TSL is very powerful in specifying complex programs, but writing TSL is hard.

To advance the usability of TSL, we developed interface designs including a structured editor with drop-down menus, a VSCode extension, and a Blockly editor. We piloted user studies and documented common challenges in writing TSL and proposed tools and strategies. We also proposed an LLM pipeline for formal logic engineering.

• Structured Editor
• TSL API
• TSL API for Animation
• Pilot Studies Blogpost

Assessing the Robustness of Electronic Health Records (EHRs)

Systems are built under the assumption that users will follow expected workflow steps. However, the actual environment may deviate — users make mistakes that can lead to unexpected, dangerous system behaviors.

To assess the robustness of EHRs, I conducted a usability study of an open-source EHR based on NIST-established guidelines and identified 40 issues. I organized the findings and proposed mitigation strategies. I also developed an automated tool, Deviated Workflow Generator, capable of generating erroneous workflows from a correct workflow, a user-specified error type, and a level of erroneous actions. It also automatically tests EHR behavior on the generated input.

• Deviated Workflow Generator

PhysioLabXR Scripting Graphic Programming Interface and Code Generation

PhysioLabXR is a Python-based open-source platform for real-time, multi-modal physiological data processing in neuroscience and HCI experiments. It's designed for researchers who aren't all programmers — but the current scripting interface requires programming knowledge.

To improve usability, I designed a graphic programming interface (GPI) for PhysioLabXR, along with a code generator that can automatically generate PhysioLabXR code from the GPI.

• GPI Design

Virtual Vitality: Augmenting Clinical Decisions via Expert-Informed Transformers

I designed an interactive interface to help clinical residents diagnose glaucoma using optical coherence tomography (OCT) reports. The system is driven by eye-tracking and a novel vision transformer (ViT) augmented by gaze behaviors from expert ophthalmologists.