1 Modeling Fatigue Predicting Performance Steven R. Hursh, Ph.D. Professor, Johns Hopkins University School of Medicine and Program Manager, Biomedical Modeling and Analysis Science Applications International Corporation, 301- 785-2341 [email protected]
Modeling Fatigue Predicting Performance. Steven R. Hursh, Ph.D. Professor, Johns Hopkins University School of Medicine and Program Manager, Biomedical Modeling and Analysis Science Applications International Corporation, 301-785-2341 [email protected]. Outline. Fatigue overview. - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
1
Modeling Fatigue Predicting Performance
Steven R. Hursh, Ph.D.
Professor, Johns Hopkins University School of Medicine
and
Program Manager, Biomedical Modeling and Analysis
Science Applications International Corporation, 301-785-2341
Fatigue overview. Drivers of fatigue Biomathematical models of fatigue and the
Sleep, Activity, Fatigue, and Task Effectiveness (SAFTE) Model
Fatigue analysis tools and the Fatigue Avoidance Scheduling Tool (FAST)
Soldier monitoring to assess fatigue Aviation applications
3
Operational Definition
Fatigue is a complex state characterized by a lack of alertness and reduced mental and physical performance, often accompanied by drowsiness.
Fatigue is more than sleepiness and its effects are more than falling asleep.
DOT Human Factors Coordinating Committee, 1998
4
Symptoms versus Root Causes
Symptoms Operational Consequences Measurable Changes in
Performance Lapses in attention and
vigilance Delayed reactions Impaired logical reasoning and
decision-making Reduced “situational
awareness” Low motivation to perform
“optional” activities Poor assessment of risk or
failure to appreciate consequences of action
Operator inefficiencies
Root Cause Analysis Fatigue is one potential root cause No direct measure, physiological marker, or “blood test” for fatigue
However, the conditions that lead to fatigue are well known and A fatigue model can help evaluation and integrate the specific conditions of an accident to determine if fatigue was involved.
5
Major Fatigue Factors
Time of Day: between midnight and 0600 hrs. Cumulative Sleep Debt: more than eight
hours accumulation. Acute Sleep Debt: less than eight hours in
last 24 hrs. Continuous Hours Awake: more than 17
hours since last major sleep period. Time on Task: continuous time doing a job
without a break.
6
Major Consequences of Fatigue
Three Mile Island (1979): 4:00 a.m. and involved human error.
Chernobyl Nuclear Reactor Meltdown (1986): 1:30 a.m. and involved human error.
Exxon Valdez (1989): 12:04 a.m. One major cause: “The failure of the third mate to properly maneuver the vessel, possibly due to fatigue and excessive workload.”
Operation Desert Storm (1990): More friendly fire losses than enemy losses, many due to sleep deprivation.
7
Benefits of Reduced Fatigue
More capable workforce – “force multiplier” Higher level of performance (higher efficiency , increased
productivity, fewer errors/incidents/accidents) Fewer accidents/incidents Reduced absenteeism, increased availability Improved health Higher moral
Time of Day Amount, quality and timing of sleep Individual factors
Phase of the circadian “pacemaker” Individual sleep need or sensitivity to
sleep loss
10
Sources of Information
Time of day: both the clock time and the time zone – inferred from location information
Sleep: Direct measurement Infer from work pattern (AutoSleep)
Duty periods and Critical Events: Drives sleep opportunities Determines critical periods for performance prediction
Individual factors Circadian phase: temperature or hormonal oscillations Sleep need: no simple test at this time
11
SAFTE
The Sleep, Activity, Fatigue, and Task Effectiveness (SAFTE) Model is based on 12 years of fatigue modeling experience and over $2.6M of US DOD investment.
Validated against laboratory and simulator measures of fatigue. Work place calibration is underway.
Now accepted by the US DOD as the common warfighter fatigue model.
Independently compared to six models from around the world and judged to have the least error (Fatigue and Performance Workshop, Seattle, 2002).
12
POC: Steven Hursh, PhD, Tel: 410-538-290112
Schematic of SAFTE™
Simulation ModelSleep, Activity, Fatigue and Task Effectiveness Model
COGNITIVEEFFECTIVENESS
SLEEP “QUALITY”FRAGMENTATION
SLEEP INTENSITY
SLEEP REGULATION
SLEEP RESERVOIR
SLEEP DEBTFEEDBACK
LOOP
INERTIA
CIRCADIAN OSCILLATORS
SLEEP ACCUMULATION(Reservoir Fill)
PERFORMANCE USE(Reservoir Depletion)
DYNAMICPHASE
PERFORMANCEMODULATION
12
13
Walter Reed Restricted Sleep Study SAFTE Model (red line) Predicts the Average Results with Precision
PVT SpeedChronic Restriction Adaptation
50
65
80
95
110
0 T1 T2 B E1 E2 E3 E4 E5 E6 E7 R1 R2 R3Day
Me
an S
pe
ed(a
s a
% o
f Ba
selin
e)
9 Hr
7 Hr
5 Hr
3 Hr
PVT SpeedChronic Restriction Adaptation
50
65
80
95
110
0 T1 T2 B E1 E2 E3 E4 E5 E6 E7 R1 R2 R3Day
Me
an
Sp
ee
d(a
s a
% o
f Ba
selin
e)
9 Hr
7 Hr
5 Hr
3 Hr
SAFTE/FAST R2 = 0.94
Restriction RecoveryBaseline
14
Driving Simulator AccidentsAccident Likelihood Relative to Baseline vs Predicted Effectiveness
y = 58.184e-0.044x
R2 = 0.8488
0
1
2
3
4
5
6
50556065707580859095100
Effectiveness (FAST Prediction)
Ac
cid
en
t L
ike
lih
oo
d (
tim
es
Ba
se
lin
e)
3 hrs sleep/day
5 hrs sleep/day
Expon. (3 hrs sleep/day)
Baseline=1(8 hrs sleep/day)
E7E6
E5
E4
E3
E2
E1
Sleep Dose Response Study (WRAIR Data)
Accident Likelihood Increases with Decreasing Effectiveness
15
Practical Software for Implementation
Fatigue Avoidance Scheduling Tool (FAST)
FAST is a fatigue assessment tool using the SAFTE model
Developed for the US Air Force and the US Army.
DOT/FRA sponsored work has lead to enhancements for
Lapse Index probably similar to values from PERCLOS drowsiness monitor.
19
Individual VariabilityDisplay Lowest 20 percentile, for example
20
BAC Scale
Arnedt, J.T., Wilde, G.J., Munt, P.W., MacLean, A.W. “How do prolonged wakefulness and alcohol compare in the decrements they produce on a simulated driving task?” Accid Anal Prev., 2001 May;33(3):337-44.
Dawson, D., Reid, K., 1997. “Fatigue, alcohol and performance impairment.” Nature 388, 23.
Continuous Hours of
Wakefulness
FAST Effectiveness
Blood Alcohol Concentration
18.5 77 0.05
21 70 0.08
Fatigue as predicted by FAST and the effects of alcohol are not identical.
The effects of fatigue may be compared to the effects of blood alcohol to calibrate the severity of fatigue
21
Dashboard InformationAnalysis System Could Report Fatigue Indicators
CriteriaValue at pointin schedule
Flags are fatigue indicators
Content based on fatigue analysis workshop hosted by NTSB and conducted by Drs. Mark Rosekind and David Dinges, funded by FRA Office of Safety.
Sleep (last 24 hrs) Chronic Sleep Debt Hours Awake Time of Day Out of Phase Performance Values
Effectiveness Mean Cognitive Lapse Index Reaction Time Reservoir
22
Sources of Uncertainty
Incomplete work/rest history, especially sleep history
Differences in personal sleep physiology Bio-rhythms Sleep need
Other personal factors Health Medications
Inaccuracies in our modeling and analysis
Lack of knowledge about specific changes in behavior
Percent of Error
Actigraphy
Temperature Sensing & GPS
Biomedical recordings
Continuous model improvement
Performance Monitoring
23
Trip Plan Editor
24
Summary of Effectiveness by Waypoints
25
Summary of Duty Periods
26
B2 Stealth Bomber
27
Lounge Chair Solution for In-flight Naps
28
Commercial Interest
Two major airlines The two largest business aviation companies Two large oil companies Five largest freight railroads A dozen electric power companies Fatigue consultants Two foreign governments
29
If you would like more information, call…………
Steven R. Hursh, Ph.D.Professor, Johns Hopkins University School of Medicine
andScience Applications International Corporation, 301-785-2341