Tariq Adnan and the PARK team received the award for “Best Poster in Engineering” at the University of Rochester’s Spring 2026 Graduate Research Day!

Tariq and the PARK team highlighted their work in tracking Parkinsonian symptoms in those that carry genes that make them more at risk to develop Parkinson’s Disease. Drawing from a global cohort, they built a multimodal framework to predict how symptoms progress over time, which is crucial for implementing early interventions.

The ROC-HCI Lab had a strong showing at the event, displaying 3 posters in total. Masum Hasan debuted his work focused on aligning LLM speech patterns with humans, called HAL. HAL is an alignment method that induces soft skills in LLMs. We applied it to medical conversations, making models more natural and human-like.  In a large-scale human study, HAL significantly outperformed both commercial and open-source models in perceived human-likeness.

Additionally, Neeley Pate showcased her work exploring whether LLMs are capable of replicating certain attributes of human likeness, such as motivated reasoning. Motivated reasoning is the idea that motivations shape how humans process information, affecting opinions and perceptions of reality. Her work finds that LLMs are not capable of replicating these human processes, suggesting further work is needed to truly emulate human decision making.

Overall, a great experience for everyone who participated in the event!