Neuropsychological test scores tap a number of underlying cognitive abilities. Examining the means by which omnibus scores are achieved provides considerable information regarding brain - behavior relationships and a richer context for clinical interpretation. This examination is the core tenant of the Boston Process Approach. Nonetheless, quantification of errors and process can be time consuming. However, the development of digital assessment technology is able to meet this challenge. For example, using a digital clock drawing test, previously unappreciated behaviors are now easily quantified and can dissociate between dementia and MCI subtypes. Research presented in this paper session provides additional insight into how digital technology can be leveraged as a powerful tool to capture behavior that, until recently, was either impractical or impossible to measure.

The assessment of graphomotor behavior can be challenging. In the context of a large-scale normative neuroimaging study, Colcombe and colleagues have engineered a digital Archimedes Spiral Test that includes measures of speed variability, rotational smoothness, and goodness of fit to the model. The temporal and spatial precision of these metrics is impressive. This research shows that, age predicted greater variable drawing speed, greater tracing errors, reduced rotational smoothness, and increasing drawing speed variability.

MacKay-Brandt and colleagues present data using a digital version of the Trail Making Tests (TMT), one of the most commonly administered neuropsychological tests. This research provides a panel of new parameters to evaluate TMT performance, including detailed speed metrics with spatial segregation to parse circle connection time from dwell time within a circle. Interestingly, dwell time, rather than traditional total time to completion, was the strongest predictor of differences between conditions and across age. Baliga and colleagues present data on a protocol of novel cancellation tests. Memory clinic patients were classified into groups presenting with mild dementia, mild cognitive impairment, and those who were cognitively normal. Digital parameters of interest included correct responses, commissions, mean intra-response latency, and mean apple pencil touch. Using these parameters, significant between group differences were obtained. Moreover, logistic regression analyses were able to classify patients into their respective groups.

It is well understood that paragraph recall tests assess a variety of underlying cognitive abilities. Andersen and colleagues studied Logical Memory recall in the Long-Life Family Study and extracted linguistic parameters that included word count, grammatical features (e.g., prepositions), and content words related to specific categories (e.g., work). Participants were classified as cognitively normal or impaired. Analyses identified distinct linguistic features of free recall that predicted cognitive status.

Hershkovich and colleagues extract measured pauses and speech frequency behavior also from a paragraph recall test. A combination of paragraph recall pause duration, speech frequency parameters, and demographic variables were able to classify older adults with and without cognitive compromise. Collectively, the evidence provided in this series of papers demonstrates that digital platforms can capture and quantify highly nuanced neurocognitive behavior to enrich information available to researchers and clinicians for analysis and clinical formulations. Digital assessment technology holds promise to realize the vision of the Boston Process Approach and revolutionize neuropsychological assessment.