RECENT PRESENTATIONS

Gevins, A. (2009). High Throughput Neurophysiological Neurocognitive Testing System. International Society for Brain Electromagnetic Topography, Kyoto, Japan.

ABSTRACT

Although cognitive brain function is affected by many diseases and medications, there is as yet no standard medical neurological assessment for it. A next-generation combined neurophysiological and neuropsychological test is introduced here. Developed with US Federal grants over the past 15 years, the test was designed as an integrated hardware/software system optimized for high-throughput automated testing of children, adults and the elderly.
An entry-level assistant administers a test in ~1 hr. EEG is recorded using a quick application headset with disposable electrodes during n-back working memory and verbal episodic memory tests, and, optionally, other tasks. Data are uploaded to the centralized analysis server where algorithms remove artifacts and compute task performance, EEG/EP parameters and multivariate functions combining performance and EEG parameters into an overall score.
Four Example Applications: During the course of development, the test has been used in research at 15 labs and clinics in the US and 1 in Japan, with >15,000 test administrations to >1500 subjects from 6 to 90 years old. Accurate and auditable automated elimination of artifacts is essential for a high-throughput system. The current 4th generation algorithms perform comparably to a human expert. On a database of ~40,000 eye-movement, head/body movement, and muscle artifacts, the algorithms detected 98% of artifacts with 3% false detections, whereas the consensus of 3 expert human judges found 96% of artifacts with 2% false detections.
1) Assessing Neurocognitive Effects of Drugs. A sensitive and efficient means for measuring a drug's neuroactivity and effects on attention and memory would be helpful to both drug developers and physicians. In 6 exploratory and 10 confirmatory studies of 6 drug types including antiepileptic, benzodiazepine, psychostimulant, antihistamine and intoxicant, Area Under the Curve (AUC) across the 16 studies was .88 (86% sensitivity, 91% specificity) when both EEG and performance measures were used, which is significantly better (p<.001) than the AUC of .76 (73% sensitivity, 77% specificity) using neuropsychological task measures by themselves. Consequently, by combining EEG and task performance measures, less than half as many subjects would be needed to assess a drug's neurocognitive effect, e.g. based on results of the 16 studies, 18 subjects would be needed for 90% power and a type I error rate of 1% using both types of measures vs. 44 subjects using task performance measures without EEG.
2) Outcome Measure for Large Scale Clinical Trials. Efficient automated test administration and uniform analysis are helpful in large scale clinical trials. A version of the test with only working memory tasks is the primary executive function outcome measure in the NIH-sponsored APPLES clinical trial of CPAP treatment of 1105 patients with obstructive sleep apnea. In that study, >99% of all 8407 tests that were administered produced valid results.
3) Tracking Early Memory Impairment in the Elderly. Tests were successfully administered to 235 elderly subjects by entry level assistants at 6 labs and clinics. 99% of all 1599 tests were valid. Retest reliability for healthy subjects across 3-, 6-, 12- and 24-month follow-ups was .84 (p<.0001). Combining memory task performance and EEG/EP measures, sensitivity of identifying impaired subjects was 83% and specificity 84% (AUC .89; p<.0001).
4) Testing Children's Attention. A version of the test for children ages 6 and up tests sustained focused and divided attention and working memory. 161 children have each been tested four times, including follow-ups at 6, 12 and 24 months, from which developmental norms are being computed. In a randomized, double-blind, cross-over, dose-response pilot study of methylphenidate (MPH) treatment of 13 children with ADHD, the test recognized the effects of MPH with 100% sensitivity and 92% specificity (AUC .96; p<.0001). A peak cognitive neurophysiological response at a particular dose was evident in each child.
Conclusion: In future prospective clinical trials, the test may prove to have usefully better sensitivity and specificity in detecting the effects of diseases and their pharmacological treatments than comparable neuropsychological test batteries that do not simultaneously measure brain function signals. If so, the test would be more effective in helping physicians optimize individual patient care. Since 3-D neurofunctional information is not essential in most medical monitoring applications, the test may also prove advantageous over analogous, but less patient friendly and more costly fMRI-based systems that may be developed. Test development has been completed and the test is available as a non-commercial research tool for Federally funded research.