Jyo S.

Research Assistant, Sci Action Canada Lab, Centre for Health Behaviour Change @Ubc Okanagan

Kelowna, BC, CA

Developed scripts to analyze 600 counseling transcripts, calculating participant–counselor speaking time ratios totest whether conversational balance predicted exercise adherence.• Applied sentiment analysis (BERT, CardiffNLP, VADER) to assess whether positive or negative tones in sessionswere associated with exercise patterns, validating findings through manual coding against the Quality ofPerformance Framework.• Implemented topic modeling to determine whether positive and negative tones reflected barriers or facilitators toexercise; cross-checked machine learning outcomes with human-coded themes.• Created data visualizations using python and participant-level reports linking transcript sentiment and dialoguepatterns with wearable data (e.g, heart rate, activity), ensuring accessibility for wheelchair users.• Contributed to knowledge translation by designing academic posters and supporting manuscripts and publishedarticles to communicate findings to both academic and applied audiences.