BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:Europe/Skopje BEGIN:DAYLIGHT DTSTART:20260329T030000 TZOFFSETFROM:+0100 TZOFFSETTO:+0200 RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3 TZNAME:CEST END:DAYLIGHT BEGIN:STANDARD DTSTART:20261025T020000 TZOFFSETFROM:+0200 TZOFFSETTO:+0100 RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=10 TZNAME:CET END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20260331T131948Z UID:75BE9968-3DFC-4ACB-812F-306E3450E2A1 DTSTART;TZID=Europe/Skopje:20260402T170000 DTEND;TZID=Europe/Skopje:20260402T183000 DESCRIPTION:Spatial omics technologies provide unprecedented maps of tissue organization\, but their complexity and heterogeneity pose major challeng es for interpretation and clinical translation. My research develops compu tational frameworks that go beyond black-box predictions by embedding biol ogical knowledge into AI systems\, creating interpretable tools that syste matically link data\, mechanisms\, and outcomes. We design specialist mode ls that capture spatial relationships\, molecular programs\, and tissue-le vel organization linking to clinically relevant observations. These models also form a foundation for biology-informed orchestrator systems capable of automatically applying methods to biological tasks\, ensuring reproduci bility\, adaptability\, and clinical relevance.\nI will present recent dev elopments in three directions. First\, relationship-based representation l earning across samples and conditions\, which captures persistent local pa tterns and provides interpretable embeddings predictive of disease progres sion and treatment response. Second\, integration approaches that augment and align data across omics modalities\, platforms\, and conditions to rev eal coherent molecular and cellular patterns\, including their trajectorie s. Finally\, graph-based modeling that associates tissue architecture with clinical outcomes\, identifying spatial features that improve patient str atification.\nThese developments are flexible\, communicable\, and readily deployable. Their outputs can be combined to define clinically relevant t issue regions\, cellular organizations\, and molecular relationships\, whi ch can then be refined with more precise experimental technologies. In thi s way\, we transform high-dimensional spatial data into mechanistic insigh t and compact spatial marker patterns and molecular panels\, accelerating the translation of discoveries into clinical practice.\n\nCo-sponsored by: FCSE\, FEEIT\, EuroCC Project\, ICT-ACT\, CyberMAC Project\n\nSpeaker(s): Jovan Tanevski\, PhD\, \n\nRoom: AMF - TMF\, Bldg: FCSE\, University Ss. Cyril and Methodius\, Faculty of Computer Science and Engineering\, Rudzer Boshkovikj 16\, Skopje\, Macedonia\, Macedonia\, 1000 LOCATION:Room: AMF - TMF\, Bldg: FCSE\, University Ss. Cyril and Methodius\ , Faculty of Computer Science and Engineering\, Rudzer Boshkovikj 16\, Sko pje\, Macedonia\, Macedonia\, 1000 ORGANIZER:katarina.trojacanec@finki.ukim.mk SEQUENCE:36 SUMMARY:Computational biomedical discovery from spatial omics data URL;VALUE=URI:https://events.vtools.ieee.org/m/551712 X-ALT-DESC:Description: <br /><p>Spatial omics technologies provide unprece dented maps of tissue organization\, but their complexity and heterogeneit y pose major challenges for interpretation and clinical translation. My re search develops computational frameworks that go beyond black-box predicti ons by embedding biological knowledge into AI systems\, creating interpret able tools that systematically link data\, mechanisms\, and outcomes. We d esign specialist models that capture spatial relationships\, molecular pro grams\, and tissue-level organization linking to clinically relevant obser vations. These models also form a foundation for biology-informed orchestr ator systems capable of automatically applying methods to biological tasks \, ensuring reproducibility\, adaptability\, and clinical relevance.<br>I will present recent developments in three directions. First\, relationship -based representation learning across samples and conditions\, which captu res persistent local patterns and provides interpretable embeddings predic tive of disease progression and treatment response. Second\, integration a pproaches that augment and align data across omics modalities\, platforms\ , and conditions to reveal coherent molecular and cellular patterns\, incl uding their trajectories. Finally\, graph-based modeling that associates t issue architecture with clinical outcomes\, identifying spatial features t hat improve patient stratification.<br>These developments are flexible\, c ommunicable\, and readily deployable. Their outputs can be combined to def ine clinically relevant tissue regions\, cellular organizations\, and mole cular relationships\, which can then be refined with more precise experime ntal technologies. In this way\, we transform high-dimensional spatial dat a into mechanistic insight and compact spatial marker patterns and molecul ar panels\, accelerating the translation of discoveries into clinical prac tice.</p> END:VEVENT END:VCALENDAR