How's my driving? GPS tracking spots Alzheimer's with 86% accuracy


Diagnosing Alzheimer's disease is still a challenge, particularly in its earliest stages, but a new study suggests that subtle changes in behaviour whilst driving could serve as an early warning system.

The researchers used GPS logging devices and machine learning as "digital biomarkers" to compare the driving of people who have preclinical Alzheimer's but are still cognitively normal to a control group of normal age-matched drivers over the course of a year.

They found that there were characteristic behaviours in the preclinical Alzheimer's group – who show subtle changes in the brain precede cognitive problems – which meant they could be distinguished from the normal controls with 86% accuracy.

The results suggest it may be possible to screen people for signs of dementia very early on, perhaps 20 years before the cognitive symptoms become apparent.

That could present a window of opportunity to try to introduce changes to lifestyle such as physical exercise or non-drug therapies like memory or cognitive training that – in theory at least – could help slow down progression to symptomatic Alzheimer's. It could potentially also help to select patients for drug therapies.

At the moment, diagnosing Alzheimer's relies on molecular biomarkers detected using PET imaging or lumbar punctures to collect cerebrospinal fluid (CSF), and access to these can be limited by cost, availability and the willingness of people to undergo an invasive procedure.

After the follow-up period, the scientists found a number of driving behaviours that were more common among the preclinical Alzheimer's group, including a tendency to take shorter trips in general and avoiding driving at night.

Other signals for preclinical Alzheimer's were driving too slowly, abrupt changes in braking or acceleration (jerking), and sticking to fewer destinations, which could imply they preferred well-known routes. The most important driving feature was jerk, according to the scientists, led by Sayeh Bayat of the Toronto Rehabilitation Institute in Canada.

GPS trackers are already being used for applications like proving careful driving patterns in order to secure cuts to insurance premiums, so it isn't a big stretch to envisage a scenario where devices are fitted for health monitoring.

The approach could provide "a non-invasive, unobtrusive, and low-cost solution" for identifying people with preclinical Alzheimer's, write the authors of the study, which is published in the journal Alzheimer's Research & Therapy.

Adding in genotype testing for ApoE4 – a well-established diagnostic biomarker for Alzheimer's pushed the accuracy of the diagnosis up, but only a little to 90%, showing that the GPS monitoring was remarkably effective on its own.

The next stage will be to carry out randomised studies of GPS monitoring that include larger numbers of subjects from a broad demographic spectrum.

Photo by Linas Drulia on Unsplash