Back

Use of Computerized Neurocognitive Assessment Software for the Detection of Alcohol Intoxication

Janarthanan, S.; Phi, H.; Flores, B.; Katz, Y.; Eagleman, D. M.; Huang, B.; Hosseini Ghomi, R.

2020-05-29 psychiatry and clinical psychology
10.1101/2020.05.12.20086868 medRxiv
Show abstract

BackgroundAcute ingestion of alcohol impairs cognitive function and poses significant threat to public health and safety with impaired operation of motor vehicles. However, there is a lack of access to tools to assess ones cognitive impairment due to alcohol. The purpose of this study was to explore the use of a neuropsychological assessment software, BrainCheck, to assess levels of alcohol impairment based on performance on the neuropsychological assessments. MethodsThe BrainCheck battery was administered to 91 volunteer participants. Participants were required to take a baseline battery prior to any alcohol ingestion, and another testing battery after a voluntary drinking period. Blood alcohol content (BAC) for the participants was obtained using a breathalyzer. Participant assessments were grouped into an intoxicated group with BAC > 0.05, and a sober group with BAC [≤] 0.05. Statistical analysis comparing intoxicated vs sober performance on the BrainCheck battery was performed. Significant metrics deemed by relative t-test and correlations with BAC were used to build statistical learning models to predict participant BAC and classify intoxicated vs. sober. ResultsRelative t-tests found two assessment metrics to be significant among comparison groups after P-value correction. Four test metrics were observed to moderately correlate (|r| > 0.40) with BAC levels. Three linear regression models (least-squares, ridge and LASSO) were built to predict participant BAC levels, with the best performing model being the least-squares model with a RMSE of 0.027, compared to the baseline RMSE of 0.061. Logistic regression models were built to classify whether the participant was intoxicated or not, with the best model performing at 80.6% accuracy, 73.3% sensitivity, and 75.0% specificity. DiscussionThe BrainCheck battery has potential to predict alcohol impairment, including participant BAC levels and if the participant is intoxicated or not. BrainCheck provides another option to assess an individuals cognitive impairment due to alcohol, with the utility of being portable and available on ones smartphone.

Matching journals

The top 12 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.