Applying mathematical models to predict resident physician performance and alertness on traditional and novel work schedules.

BACKGROUND: In 2011 the U.S. Accreditation Council for Graduate Medical Education began limiting first year resident physicians (interns) to shifts of ≤16 consecutive hours. Controversy persists regarding the effectiveness of this policy for reducing errors and accidents while promoting education and patient care. Using a mathematical model of the effects of circadian rhythms and length of time awake on objective performance and subjective alertness, we quantitatively compared predictions for traditional intern schedules to those that limit work to ≤ 16 consecutive hours. METHODS: We simulated two traditional schedules and three novel schedules using the mathematical model. The traditional schedules had extended duration work shifts (≥24 h) with overnight work shifts every second shift (including every third night, Q3) or every third shift (including every fourth night, Q4) night; the novel schedules had two different cross-cover (XC) night team schedules (XC-V1 and XC-V2) and a Rapid Cycle Rotation (RCR) schedule. Predicted objective performance and subjective alertness for each work shift were computed for each individual's schedule within a team and then combined for the team as a whole. Our primary outcome was the amount of time within a work shift during which a team's model-predicted objective performance and subjective alertness were lower than that expected after 16 or 24 h of continuous wake in an otherwise rested individual. RESULTS: The model predicted fewer hours with poor performance and alertness, especially during night-time work hours, for all three novel schedules than for either the traditional Q3 or Q4 schedules. CONCLUSIONS: Three proposed schedules that eliminate extended shifts may improve performance and alertness compared with traditional Q3 or Q4 schedules. Predicted times of worse performance and alertness were at night, which is also a time when supervision of trainees is lower. Mathematical modeling provides a quantitative comparison approach with potential to aid residency programs in schedule analysis and redesign.

Human fatigue and the crash of the airship Italia.

The airship Italia, commanded by General Umberto Nobile, crashed during its return flight from the North Pole in 1928. The cause of the accident was never satisfactorily explained. We present evidence that the crash may have been fatigue-related. Nobile's memoirs indicate that at the time of the crash he had been awake for at least 72 h. Sleep deprivation impairs multiple aspects of cognitive functioning necessary for exploration missions. Just prior to the crash, Nobile made three command errors, all of which are of types associated with inadequate sleep. First, he ordered a release of lift gas when he should have restarted engines (an example of incorrect data synthesis, with deterioration of divergent thinking); second, he inappropriately ordered the ship above the cloud layer (a deficiency in the assessment of relative risks); and third, he remained above the cloud layer for a prolonged period of time (examples of attention to secondary problems, and calculation problems). We argue that as a result of these three errors, which would not be expected from such an experienced commander, there was no longer enough static lift to maintain level flight when the ship went below the cloud layer. Applying Circadian Performance Simulation Software to the sleep-wake patterns described by Nobile in his memoirs, we found that the predicted performance for someone awake as long as he had been is extremely low. This supports the historical evidence that human fatigue contributed to the crash of the Italia.

On-line EEG Denoising and Cleaning Using Correlated Sparse Recovery and Active Learning.

We have developed two new methods that use sparse recovery and active learning techniques for near real-time artifact identification and removal in EEG recordings. The first algorithm, called Correlated Sparse Signal Recovery (CSSR) addresses the problem of structured sparse signal recovery when statistical rather than exact properties describing the structure of the signal are appropriate, as in the elimination of eye movement artifacts; such tasks cannot be done efficiently using structured models that assume a common sparsity profile of fixed groups of components. Our algorithm learns structured sparse coefficients in a Bayesian paradigm. Using it, we have successfully identified and subtracted eye movement (structured) artifacts in real EEG recordings resulting in minimal data loss. Our method outperforms ICA and standard sparse recovery algorithms by preserving both spectral and complexity properties of the denoised EEG. Our second method uses a new active selection algorithm that we call Output-based Active Selection (OAS). When applied to the task of detection of EEG epochs containing other non-structured artifacts from an ensemble of detectors, OAS boosts accuracy of the ensemble from 91% to 97.5% with only 10% active labels. Our methods can also be applied to real-time artifact removal in magnetoencephalography (MEG) and blood pressure signals.