Project Grant R44GM154526

HARNESSING MACHINE LEARNING & OBJECT DETECTION FOR AUTOMATED EVALUATION OF STUDENT-FOLDED PROTEIN MODELS - PROJECT SUMMARY HARNESSING MACHINE LEARNING & OBJECT DETECTION FOR AUTOMATED EVALUATION OF STUDENT-FOLDED PROTEIN MODELS MODELING IS A KEY PRACTICE IN SCIENCE EDUCATION. MINI-TOOBERS (FOAM-COVERED WIRE) ARE ONE OF 3D MOLECULAR DESIGNS' (3DMD) MOST INNOVATIVE, POPULAR, AND EFFECTIVE MODELING MATERIALS, MOST USED TO REPRESENT PROTEIN STRUCTURES IN BIOLOGY CLASSROOMS AND OUTREACH PROGRAMS NATIONWIDE. AS A FREE-FORM MODELING MATERIAL, MINI-TOOBERS CAN BE FOLDED INTO AN ENDLESS VARIETY OF SHAPES AND FORMS. HOWEVER, DUE TO THEIR FREE-FORM NATURE, MINI-TOOBERS LACK AN IMMEDIATE FEEDBACK SYSTEM, MAKING THEM DIFFICULT TO USE EFFICIENTLY IN FORMAL AND INFORMAL SCIENCE EDUCATION SETTINGS. 3DMD'S MORE RIGID MODELING TOOLS HAVE BUILT-IN FEEDBACK FEATURES LIKE MAGNETS, ARROWS, POST/HOLE CONNECTIONS, ETC. THAT HELP STUDENTS SELF-CORRECT AND ALLOW EDUCATORS TO MAKE FORMATIVE ASSESSMENTS QUICKLY. HOWEVER, THERE IS OFTEN TOO MUCH TO OBSERVE IN A MINI-TOOBER MODEL (SHAPE, SIDECHAINS, SECONDARY STRUCTURES, ETC.) FOR AN EDUCATOR TO ASSESS ALL THE DATA IN REAL TIME EFFECTIVELY. THE SOLUTION IS TO USE MACHINE LEARNING (ML) AND OBJECT DETECTION (OD) TECHNOLOGIES TO ANALYZE MINI-TOOBER PROTEIN MODELS AND PROVIDE STUDENTS AND TEACHERS WITH IMMEDIATE DIGITAL FEEDBACK AND PRECISE ASSESSMENT OF THEIR MODELS. THE PROJECT AIMS TO 1) DEVELOP AND PUBLISH SMARTPHONE APPLICATIONS FOR STUDENT TRAINING AND EDUCATOR ASSESSMENT THAT USE MACHINE LEARNING AND OBJECT DETECTION TO PROVIDE FEEDBACK ON FREE-FORM-MEDIA-MODELED PROTEINS. 2) EVALUATE THE STUDENT TRAINING AND EDUCATOR ASSESSMENT SMARTPHONE APPLICATIONS TO ASSESS USABILITY, ACCURACY & LEARNING GAINS. ACHIEVING THESE AIMS WILL HELP IMPROVE BIOSCIENCE EDUCATION IN SECONDARY SCHOOLS, RESULTING IN AN EXPANDED STEM WORKFORCE AND A SCIENTIFICALLY LITERATE CITIZENRY. IN ADDITION, THIS PROJECT MOVES THE FIELD OF IMMERSIVE TECHNOLOGIES FORWARD BY DEVELOPING NOVEL APPROACHES TO OD & ML SYSTEMS TO RECOGNIZE AND EVALUATE FREE-FORM MEDIA. TWO APPLICATIONS WILL BE CREATED: 1) PROTEIN MODELING STUDENT TRAINING APP WILL INTRODUCE PROTEIN STRUCTURE-FUNCTION CONCEPTS AND HOW TO FOLD MINI-TOOBER PROTEIN MODELS. THE OD SYSTEM WILL PROVIDE REAL-TIME DIGITAL FEEDBACK TO INFORM PROPER MODELING TECHNIQUES. 2) PROTEIN MODELING EDUCATOR ASSESSMENT APP WILL QUICKLY AND ACCURATELY SCORE FOLDED MINI-TOOBER PROTEIN MODELS. THE OD SYSTEM WILL IDENTIFY PROTEIN MODEL COMPONENTS, ANALYZE THEIR RELATIVE POSITIONS, AND PROVIDE ACCURACY SCORES FOLLOWING A STANDARD PROTEIN JUDGING RUBRIC. THE TWO APPLICATIONS WILL BE DEPLOYED THROUGH THE APPLE APP AND GOOGLE PLAY STORES FOR STUDENTS AND EVENT SUPERVISORS PARTICIPATING IN THE SCIENCE OLYMPIAD PROTEIN MODELING EVENT, AN ANNUAL SCIENCE COMPETITION INVOLVING 6,000 HIGH SCHOOL TEAMS NATIONWIDE. THE APPLICATIONS' ABILITY TO ANALYZE MINI-TOOBER PROTEIN MODELS AND PROVIDE REAL-TIME FORMATIVE FEEDBACK WILL BE ASSESSED FOR USABILITY, ACCURACY OF THE ML & OD SYSTEM, AND EDUCATIONAL LEARNING GAINS.