Development and validation of novel tests in the DCT drawing analysis platform for the detection of Alzheimer’s Disease-related early cognitive

项目总结/文摘

Project Summary/Abstract Early detection of AD and AD-related dementias is critical for the development of novel therapeutic agents and for effective patient treatment and care. Linus Health has acquired a platform able to detect subtle changes in behavior indicative of early cognitive impairment by using a digital pen to record drawing motions and automatically analyzing that data using a combination of artificial intelligence and neuropsychological analysis. DCTclockTM, the first test on this platform, has been successfully launched and is able to detect subtle cognitive change in early Alzheimer’s Disease (AD) and AD-related dementias earlier than standard cognitive tests, correlate with known AD biomarkers such as amyloid and tau in preclinical-AD subjects, and detect known APOE genetic risk in cognitively healthy participants. The test is FDA-cleared to market for general cognitive assessment and is currently used in multinational clinical trials, research initiatives, and by practicing clinicians for patient care. Linus Health is now conducting a study involving ~200 participants ranging from cognitively healthy to demented to validate a tablet as input device for the drawing data, and to collect data on a set of novel tablet-based drawing tasks including: a symbol task, a visual retention task, a trail making task, a path-finding task, and a visuoconstructional task. We aim to analyze task data collected in the study to produce a set of easily-deployable, rapid, cost-effective, and sensitive tests to be used as cognitive digital biomarkers in clinical trials for AD and AD-related dementias. The first aim is to develop novel cognitive measures from these additional tests and obtain preliminary validation. Existing data will be split into training and testing sets. The training set will be analyzed to develop metrics, following an interpretable hierarchical analysis structure with low-level drawing measurements combined in a set of composite scales tied to a cognitive concept (i.e., information processing), and then further combined to obtain a simple 0-100 score for performance on the test. The testing set will be used to conduct preliminary validation consisting of a Receiver Operating Characteristic curve analysis, correlation measures to existing neuropsychological tests, and test-retest reliability measurements to ensure repeat testing stability. The second aim is to implement the developed metrics in the commercial Linus platform. Robust software will be created for the measurement, storage, and display of the novel metrics, setting a foundation to enable usage of the metrics by research and commercial partners.