P30AG103094↗

DUKE ARTIFICIAL INTELLIGENCE AND TECHNOLOGY COLLABORATORY FOR AGING: GOVERNANCE AND EVALUATION (DUKE AITC-AIGE) - PROJECT SUMMARY: OVERALL THE DUKE ARTIFICIAL INTELLIGENCE AND TECHNOLOGY COLLABORATORY FOR AGING: GOVERNANCE AND EVALUATION (DUKE AITCAIGE) IS A NATIONAL INITIATIVE TO BRIDGE THE IMPLEMENTATION GAP BETWEEN ARTIFICIAL INTELLIGENCE (AI) RESEARCH AND ITS APPLICATION IN CARE FOR OLDER ADULTS AND INDIVIDUALS WITH ALZHEIMER'S DISEASE AND RELATED DEMENTIAS (AD/ADRD). THE PROGRAM DIRECTLY ADDRESSES FIVE CRITICAL BARRIERS TO PROGRESS: A LACK OF STAKEHOLDER INPUT IN PROBLEM DEFINITION, DATA LIMITATIONS IN TRAINING DATA, INADEQUATE VALIDATION OF SUBGROUP PERFORMANCE, INSUFFICIENT MONITORING FOR MODEL DRIFT, AND UNACCOUNTABLE GOVERNANCE. ITS PRIMARY OBJECTIVE IS TO CREATE A NATIONAL ECOSYSTEM THAT ACCELERATES THE DEVELOPMENT AND IMPLEMENTATION OF TRUSTWORTHY, RESPONSIBLE, AND PERSON-CENTERED AI. THE COLLABORATORY'S ORGANIZATIONAL FRAMEWORK INTEGRATES THREE SYNERGISTIC CORES. THE ADMINISTRATIVE CORE (AC) PROVIDES NATIONAL LEADERSHIP AND ETHICAL OVERSIGHT, MANAGES A NATIONAL PILOT PROJECT COMPETITION, AND SERVES AS THE PRIMARY LIAISON WITH THE AITC COORDINATING CENTER (AITCC) TO ENSURE NATIONAL SYNERGY. THE RESEARCH INNOVATION CORE (RIC) SERVES AS A DISCOVERY ENGINE BY FUNDING AND MENTORING INTERDISCIPLINARY PILOT PROJECTS. THE CLINICAL IMPLEMENTATION AND DATA SCIENCE (CIDS) CORE OFFERS SPECIALIZED CONSULTATION FOR TRANSLATION, UTILIZING THE DYNAMIC DATA RELIABILITY (DDR) FRAMEWORK TO QUANTIFY DATA ‘FIT-FOR-USE’ AND ENSURE DATASETS ARE AIREADY. A CENTRAL INNOVATION IS THE TRUSTWORTHY AI IMPLEMENTATION AND GOVERNANCE (TRIG) FRAMEWORK, A MANDATORY METHODOLOGY FOR ALL PROJECTS THAT OPERATIONALIZES FAIRNESS, ACCOUNTABILITY, AND TRANSPARENCY INTO REPRODUCIBLE METHODS. THE TRIG FRAMEWORK REQUIRES CONCRETE ACTIONS, INCLUDING FAIRNESS AUDITS (E.G., DEMOGRAPHIC PARITY, EQUALIZED ODDS), CONTINUOUS MONITORING FOR PERFORMANCE DRIFT, AND TRANSPARENT DOCUMENTATION VIA MODEL CARDS. IT INCORPORATES ADVANCED METHODS LIKE METAMORPHIC TESTING FOR MODEL ROBUSTNESS AND MANDATORY SECURITY AND PRIVACY VETTING TO ASSESS RESILIENCE AGAINST ADVERSARIAL ATTACKS AND PROMOTE PRIVACY-PRESERVING TECHNIQUES LIKE FEDERATED LEARNING. THIS SOCIOTECHNICAL MODEL IS FOUNDED UPON DUKE'S ESTABLISHED STRENGTHS IN GERIATRICS (PEPPER CENTER), AD/ADRD RESEARCH (DUKE/UNC ADRC), TRANSLATIONAL SCIENCE (DUKE CTSI), AND AI GOVERNANCE (DUKE AI HEALTH). THIS STRUCTURE CREATES AN INTEGRATED FEEDBACK LOOP WHERE REAL-WORLD CLINICAL AND COMMUNITY NEEDS DRIVE TECHNOLOGY DEVELOPMENT AND VALIDATION. AN EXEMPLAR APPLICATION IS IN POLYPHARMACY MANAGEMENT FOR AD/ADRD, WHERE TRIG GUIDES AI TOOL DEVELOPMENT CO-DESIGNED WITH PATIENTS AND CLINICIANS TO REDUCE INAPPROPRIATE PRESCRIBING. THE INTENDED NATIONAL IMPACT IS TO ESTABLISH A NEW STANDARD FOR TRANSLATIONAL AI SCIENCE BY CREATING A SUSTAINABLE PIPELINE OF RIGOROUSLY VALIDATED AND ETHICALLY GOVERNED AI INNOVATIONS. BY YEAR 5, THE AITC-AIGE WILL DISSEMINATE REPRODUCIBLE GOVERNANCE RESOURCES, CHAIALIGNED REPORTING CHECKLISTS, AND A NATIONAL SAFETY REPORTING INFRASTRUCTURE. THESE EFFORTS WILL ADVANCE THE NIA'S MISSION AND ENSURE AI DELIVERS GENERALIZABLE IMPROVEMENTS IN INDEPENDENCE AND QUALITY OF LIFE FOR OLDER ADULTS.

Project Grant (B) · National Institutes of Health · to Duke University (UEI TP7EK8DZV6N5) · data as of 2026-09-30

Source: USAspending.gov prime award transactions and FSRS subaward reports, as loaded by SpendQuery (data as of 2026-09-30). Amounts are obligations (money committed), not outlays. The official record on USAspending.gov ↗