Digital biomarkers in narrative and conversational speech in Frontotemporal Dementia

今天，临床标准是识别额颞叶痴呆（FTD）变体的金标准。虽然好

Today clinical criteria are the gold standard to identify variants of Frontotemporal Dementia (FTD). Though well established and useful clinically, these criteria rely on lengthy procedures that are expensive and invasive. There is urgent need for reliable, objective, and valid measures to screen for likely pathology and monitor disease longitudinally in disease-modifying treatment trials. While preparing the PI to become an independent translational researcher, this research program addresses clinical objectives to develop novel, reproducible, inexpensive, non-invasive, validated biomarkers derived from automated analyses of digitized speech in patients with Frontotemporal Lobar Degeneration (FTLD) pathology. This program is also designed with the scientific objectives to enrich neurobiologic models of language with essential but missing speech components and to improve our pathophysiologic understanding of spreading pathology. In Aim 1 of the K99 mentored phase, the PI will identify informative digitized acoustic and lexical markers in antemortem speech and relate these directly to regional pathologic burden in FTLD-Tau and FTLD-TDP, and uniquely study these speech markers and associated MRI cortical thinning in presymptomatic and symptomatic mutation carriers. In Aim 2, she will examine acoustic and lexical markers longitudinally in FTLD-Tau and FTLD-TDP and in mutation carriers. The PI’s preliminary data show that automated algorithms such as automated speech recognition (ASR) and natural language processing (NLP) tools extract novel digital markers from speech that are sensitive and specific to FTD phenotypes, are associated with imaging and biofluid markers of underlying FTLD pathology, and are related directly to regional pathologic burden in cross-sectional and longitudinal studies of autopsied FTLD-Tau and FTLD-TDP as well as carriers of mutations related to FTLD. During the R00 independent phase, Aim 3 will discover novel markers in the highly elusive domain of natural conversational speech, the most common form of human-to-human communication, and relate these to biomarkers of FTD. The PI will accomplish these aims with the guidance of a world-class mentoring team (Drs. Murray Grossman, Mark Liberman, James Gee, and David Irwin) who have expertise in computational linguistics, neuroimaging, and experimental neuropathology. The PI has available a unique database of 1500 digitized speech samples in deeply endophenotyped autopsied cases and mutation carriers. The mentoring team has a very strong track record of training postdoctoral fellows for independent research careers. A detailed training plan includes regular one-on-one meetings, seminars, coursework, grant writing, and responsible conduct of research. The new skills acquired during this period, combined with the PI’s prior expertise in cognitive neurology and leadership experience as a Major in the military, will ensure a strong foundation to launch an independent translational research career that will synergistically address vital scientific and clinical issues.