BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/Chicago BEGIN:DAYLIGHT DTSTART:20260308T030000 TZOFFSETFROM:-0600 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:CDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20251102T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0600 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:CST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20251118T025146Z UID:A3F55D40-15F9-4055-A8C5-E8F47A43DA76 DTSTART;TZID=America/Chicago:20251117T193000 DTEND;TZID=America/Chicago:20251117T203000 DESCRIPTION:Abstract:\nReading assessments are essential for enhancing stud ents' comprehension\; yet\, many EdTech applications focus mainly on outco me-based metrics\, providing limited insights into students' reading behav iors and cognition. This study investigates the use of multimodal data tha t includes eye-tracking data\, along with learning outcomes\, assessment c ontent\, and teaching standards to derive meaningful reading insights. We employ unsupervised learning techniques to identify distinct reading behav ior patterns. A large language model (LLM) then synthesizes the derived in formation into actionable reports for educators\, streamlining the interpr etation process. LLM experts and human educators evaluated these reports f or clarity\, accuracy\, relevance\, and pedagogical usefulness. Our findin gs indicate that LLMs can effectively function as educational analysts\, t urning diverse data into teacher-friendly insights that educators find ben eficial. While automated insight generation shows promise\, human oversigh t remains crucial to ensure reliability and fairness. This research advanc es human-centered AI in education\, connecting data-driven analytics with practical classroom applications.\n\nBio: Dr. Eduardo Davalos is an Assist ant Professor at Trinity University\, working at the intersection of AI in Education (AIED)\, Human–Computer Interaction (HCI)\, and Large Languag e Models (LLMs). His research develops privacy‑preserving\, browser‑na tive sensing and modeling techniques that translate into scalable learning technologies. His PhD is in Computer Science from Vanderbilt University\, where he and my team developed RedForest\, a e-learning platform that inc orporates AI to assist teacher workflows\, including assessment creation a nd other aspects such as gaze analytics and collaborate learning/play. His latest focus is on finding ways to incorporate AI agents to meaningfully assist teachers and students by providing more personalize feedback\, sugg estions\, and content.\n\nRoom: BSIC 203 - Data Science and Machine Learni ng Lab\, Bldg: Blank Sheppard Innovation Center (next to building #22 in t he map)\, One Camino Santa Maria\, St. Mary's University of San Antonio\, San Antonio\, Texas\, United States\, 78228 LOCATION:Room: BSIC 203 - Data Science and Machine Learning Lab\, Bldg: Bla nk Sheppard Innovation Center (next to building #22 in the map)\, One Cami no Santa Maria\, St. Mary's University of San Antonio\, San Antonio\, Texa s\, United States\, 78228 ORGANIZER:wluo@stmarytx.edu SEQUENCE:5 SUMMARY:LLMs as Educational Analysts: Transforming Multimodal Data Traces i nto Actionable Reading Assessment Reports URL;VALUE=URI:https://events.vtools.ieee.org/m/499770 X-ALT-DESC:Description: <br /><div><strong data-olk-copy-source="MessageBod y">Abstract:</strong>&nbsp\;\n<div data-olk-copy-source="MessageBody"><spa n data-olk-copy-source="MessageBody">Reading assessments are essential for enhancing students' comprehension\; yet\, many EdTech applications focus mainly on outcome-based metrics\, providing limited insights into students ' reading behaviors and cognition. This study investigates the use of mult imodal data that includes eye-tracking data\, along with learning outcomes \, assessment content\, and teaching standards to derive meaningful readin g insights. We employ unsupervised learning techniques to identify distinc t reading behavior patterns. A large language model (LLM) then synthesizes the derived information into actionable reports for educators\, streamlin ing the interpretation process. LLM experts and human educators evaluated these reports for clarity\, accuracy\, relevance\, and pedagogical usefuln ess. Our findings indicate that LLMs can effectively function as education al analysts\, turning diverse data into teacher-friendly insights that edu cators find beneficial. While automated insight generation shows promise\, human oversight remains crucial to ensure reliability and fairness. This research advances human-centered AI in education\, connecting data-driven analytics with practical classroom applications.</span></div>\n</div>\n<di v>&nbsp\;</div>\n<div><strong>Bio:</strong> Dr. <span data-olk-copy-source ="MessageBody">Eduardo Davalos is</span><span data-olk-copy-source="Messag eBody"> an Assistant Professor at Trinity University\, working at the inte rsection of AI in Education (AIED)\, Human&ndash\;Computer Interaction (HC I)\, and Large Language Models (LLMs). His research develops privacy‑pre serving\, browser‑native sensing and modeling techniques that translate into scalable learning technologies. His PhD is in Computer Science from V anderbilt University\, where he and my team developed RedForest\, a e-lear ning platform that incorporates AI to assist teacher workflows\, including assessment creation and other aspects such as gaze analytics and collabor ate learning/play. His latest focus is on finding ways to incorporate AI a gents to meaningfully assist teachers and students by providing more perso nalize feedback\, suggestions\, and content.</span></div> END:VEVENT END:VCALENDAR