The Digital-Physical Intersection
Subtle changes in behaviour may warn of early-onset of Parkinson’s Disease before it becomes diagnosable. One question is how to measure this? PDLab have published research detecting PD using machine learning to analyse users’ typing pattern, which is a pretty clever, unobtrusive source of data.
Similarly, this research demonstrates that early-onset Alzheimers can be seen through changes in writing style (example table below taken from paper).
|Linguistic marker||Normal aging||Dementia|
|Vocabulary size||Gradual increase, possible slight decrease in later years||Sharp decrease|
|Repetition||Possible slight decrease/increase||Pronounced increase|
|Word specificity||Possible slight increase/decrease||Pronounced decrease|
|Word class deficit||Insignificant change||Pronounced deficit in nouns; possible compensation in verbs|
|Fillers||Possible slight increase||Pronounced increase|
|Overall complexity||No change or gradual decline, possible rapid decline around mid-70s||Sharp decline|
|Use of passive||Possible slight decrease||Pronounced decrease|
|Auxiliary verb||Be-passives dominate||Get-passives dominate|
|Agentless passive||Moderate decrease||Greater decrease|
There’s also an interesting radiolab podcast on this latter research.