University Of California-Los Angeles receives $169K Aging Research grant from the National Institutes of Health

Advancing early detection of mild cognitive impairment and dementia through digital biomarkers - project summary/abstract this mentored patient-oriented research career development award (K23) will provide protected time for dr. Tue te to develop a focused program of research on early detection of mild cognitive impairment (mci) in older adults. The aims of this 5-year career development plan will enhance dr. Te’s knowledge and skills in: 1) mixed methods research to explore patient and provider perspectives on voice-based cognitive screening tools in real-world clinical settings, 2) voice signal processing and machine learning for developing digital biomarkers for early detection of mci and alzheimer’s disease and related dementias (ad/adrd), 3) clinical and research expertise in the diagnosis and management of ad/adrd, and 4) grant writing to support transition to research independence. The career development plan includes structured mentorship meetings, coursework and workshops, and dissemination of research findings through peer-reviewed publications and conference presentations. Research activities during the K23 award include conducting the K23 study (e.g., focus groups on patient and clinician perspectives, development and validation of a voice preprocessing pipeline accounting for linguistic complexity, and construction of predictive machine learning models for mci detection). Findings will be disseminated in multiple manuscripts submitted to peer-reviewed journals and conferences. The proposed research uses a multimodal, longitudinal approach integrating voice signal processing, sleep measures (including single-channel eeg over multiple nights), blood/fluid biomarkers, and electronic medical records to address key gaps in mci detection. Specific aims are to: 1) assess the feasibility and acceptability of collecting and analyzing spontaneous voice data during outpatient visits, 2) develop machine learning models using voice/speech features to distinguish mci from normal cognition while controlling for demographics, comorbidities, biomarkers, and sleep measures, and 3) explore associations between voice features and objective (nrem slow wave activity) and subjective (epworth sleepiness scale ≥10) sleepiness, testing whether voice-based measures predict sleepiness comparable to eeg-based assessments. Dr. Te’s multidisciplinary mentoring team includes primary mentor constance fung, an NIH-funded sleep and aging researcher, and co-mentors drs. Brendan lucey (ad/adrd and sleep), derjung mimi tarn (qualitative research), alex a.t. Bui (voice signal processing and ai), and mary regina boland (biomedical informatics). Mentors will provide guidance through structured meetings, collaborative research, and skill development in manuscript and grant writing, ensuring dr. Te’s successful transition to research independence. This K23 project will lay the groundwork for integrated voice–sleep models reflecting real-world complexity, supporting low-burden, scalable tools for early detection of ad/adrd in outpatient settings, and facilitating future R01 applications to advance preclinical prediction of cognitive impairment.