Cognitive Assessment using Mobile Phones

Everyday Assessment of the Effects of Alcohol

Cognitive assessments, such as tests of memory, attention, and reaction time can be set up on mobile phones. This has important advantages:

  • Assessments can be made in an everyday life setting - home or work - and the unfamiliar and artificial surroundings of the lab avoided.
  • Assessments can be made frequently, and at different times during the day, and a profile of the person's performance built up
  • Links between different aspects of a person's life can be studied. If you miss breakfast does it have an effect on your memory and concentration later in the day? What about if you are stressed, or didn't sleep well last night. Or maybe you have had a few drinks. How much does that really affect you?
  • Data can be transmitted over the network after each assessment, so that data are promptly available for review, 

We carried out a study to evaluate this method, looking at the effects of alcohol consumption. We know from lab studies that alcohol affects performance on a range of tests. We set up a selection of tests on the mobile phone that we expected to be affected by alcohol. Some tests are not well adapted to the small screen on a mobile, but there are plenty of tests that will easily fit. Here is an example:

Number Pairs Test

Five digits appear on the screen. The task is to check if the second and fourth digits are the same, and to press a YES button if they are the same, a NO button if they are different

In the example on the left, both the second and fourth digits are 2, and so the correct response is Yes.

This is a speeded test, so volunteers are asked to respond as quickly and as accurately as they can. Reaction times and correct and incorrect responses are recorded.

We recruited 38 volunteers. In this study we gave them phones to use, though in more recent studies volunteers have used their own phones. They carried out assessments twice a day for two weeks. We sent them an SMS message each time we wanted them to complete the tests. The times were randomised so that we would get a good spread of completion times.

Each assessment started with questions about how much alcohol they had consumed in the past 6 h and that day. We used this to divide our assessments into 3 groups:

  • No alcohol (NA): None consumed during the current day
  • Significant Alcohol (SA): 5 or more units (1 unit = 8g pure alcohol) in the previous 6 h.
  • Intermediate

The main analysis compared SA to NA. For the Number-Pairs test, accuracy (proportion of correct responses) after SA was worse than after NA (comparisons were time-matched). On average the were 10.5%  incorrect responses for SA  and 4.4 for NA. Performance on the two other tests we used was also impaired

After the two weeks of everyday assessment, volunteers took part in a lab study, looking at the same tests, with alcohol given as a drink containing vodka, which was compared to a matching drink containing no vodka. Doses were calculated to give blood alcohol a bit over the UK legal driving limit. The study used the pharmacological model, where the amount given was pre-determined, the drink had to be consumed in a short time, and dosage was double-blind. In short the typical artificial laboratory situation. In these lab tests, alcohol impaired performance, in particular increasing errors in number pairs from 4.7% to 7.3%. So the effect was a bit smaller in the lab. This was true of the other tests as well.

We did not have a direct measurement of blood alcohol in the everyday part of the study. We estimated this from a previous pub study where we asked drinkers how many drinks they had taken, and also breathalysed them. Using this relationship, we calculated that blood alcohol in our everyday assessments with SA was a bit lower than in the lab sessions. If our estimates are correct, we had a larger effect in everyday assessment with a lower blood alcohol. So while the lab and everyday assessments agree about the type of impairment we see, if anything, the lab studies underestimate the size of the impairment.

The other conclusion from the study was that everyday assessment using mobile phones was a practicable and effective method of measuring cognitive function..

The full reference to the paper is

Tiplady, B., Oshinowo, B., Thomson, J., & Drummond, G. B. (2009) Alcohol and Cognitive Function: Assessment in Everyday Life and Laboratory Settings Using Mobile Phones, Alcoholism Clinical and Experimental Research 33(12) 412-418. Abstract