Why Four Minutes? The Case for Brief, Daily Assessment

A comprehensive neuropsychological evaluation takes 6-8 hours and produces detailed, rich data about cognitive function. It is also done once every 12-24 months at best, by most clinical protocols. That is 6 data points per decade — not nearly enough to detect the early, gradual change that characterizes prodromal neurodegeneration.

A four-minute daily session produces 365 data points per year, each less detailed but collectively revealing a longitudinal picture that a 6-hour annual test cannot match. For the purpose of detecting personal change over time, frequency beats depth. This is a well-established principle in psychometrics — the reliability of a measure increases as a function of the number of observations, even when individual observations are noisy.

The feasibility of brief, repeated cognitive monitoring has been established in multiple research contexts. The Lumosity Neuroscience lab published data in 2014 showing that brief (10-minute) daily cognitive assessments, taken consistently for months, were sensitive to age-related change and health status. The Alzheimer's Association has funded research into digital biomarkers — using brief daily tasks on consumer devices to detect early cognitive change.

A 2022 study in Alzheimer's and Dementia: Diagnosis, Assessment and Disease Monitoring found that self-administered digital cognitive assessments, when taken frequently (daily or near-daily), produced data comparable to clinician-administered neuropsychological tests for detecting longitudinal change — even though any single brief session had lower sensitivity than a full clinical evaluation.

The critical variable is not session length. It is the consistency and frequency of measurement.

Clinical researchers have known for decades that assessment burden is the primary driver of dropout in longitudinal cognitive studies. Participants who complete 1-hour monthly assessments drop out at much higher rates than those completing 10-minute weekly ones, even when controlling for population characteristics. The data you never collect because the participant quit the study is worse than imperfect data from a brief test.

In consumer health apps, this effect is even more pronounced. Keel's four-minute format was chosen based on user research showing that four minutes sits at the threshold between 'I can do this every day' and 'this is a commitment I'll skip.' Below four minutes, the battery cannot sample the range of cognitive domains worth tracking. Above six minutes, daily compliance drops sharply.

Cognitive decline is domain-specific in its early stages. Alzheimer's disease typically impairs episodic and semantic memory before affecting processing speed. Vascular cognitive impairment typically impairs processing speed and executive function before memory. Frontotemporal dementia impairs executive function and semantic memory in a different profile than Alzheimer's.

A single measure — even an excellent one — cannot distinguish these profiles. Keel's five tests cover five distinct domains (processing speed, visuospatial working memory, semantic fluency, attentional processing, and working memory), giving enough coverage to detect domain-specific patterns. Within four minutes, this is the minimum viable battery for meaningful longitudinal cognitive monitoring.