1. STOP WATCH TESTING
An analysis into the use of a new testing technique to
identify and treat concussions
By Peter Eggleston Connor
Data provided by
Southern Oregon Orthopedics
A Graduate Thesis from
Southern Oregon University
3. HISTORY
• Prior to 2001: Concussion detection had little empirical evidence
supporting it.
• 2001-2012: American Academy of Neurology (AAN) creates guidelines
for more globally accessing concussion risks in athletes.
• 2013: An update came out pointing towards evidence that having a
concussion made future concussions more likely.
4. OBJECTIVES
• Identify changes between healthy and injured reaction times
• Determine a concussion recovery rate based on reaction time
• Identify significant symptom scores over course of recovery
9. DESCRIPTIVE ANALYSIS – RECOVERY TIME
n = 39
Mean:
8.2 days
Standard Deviation:
5.3 days
90th Percentile:
~11 days
10. DESCRIPTIVE ANALYSIS – HEALTHY REACTION TIME
Mean:
0.18 sec
Standard Deviation:
0.03 sec
11. DESCRIPTIVE ANALYSIS –
HEALTHY REACTION TIME IN CONCUSSED PLAYERS
Mean:
0.18 sec
Standard Deviation:
0.02 sec
Hypothesis Test All vs
Concussed Healthy
Reaction Times
p-value:
0.16
Conclusions:
Accept Null
12. DESCRIPTIVE ANALYSIS – INJURED REACTION TIME
Mean:
0.27sec
Standard Deviation:
0.10sec
Hypothesis Test Healthy vs
Injured Reaction Time in
Concussed Athletes
p-value:
~1
Conclusion:
Reject Null
16. EXPONENTIAL
MODELING:
DIFFERENCE 𝑦 = (𝑐 − ℎ) 𝑒 𝛽𝑡
+ ℎ
y: The reaction time at t days since injury
c: The reaction time at the time of injury
h: The healthy reaction time
β: The rate of decay in the reaction time during recovery
t: The number of days since the concussion injury
21. PREDICTIVE MODEL USING DIFFERENCES
Can be transformed into:
𝑡 𝑦,𝑖 =
𝑙𝑜𝑔(
𝑘
𝑐𝑖 − ℎ𝑖
−0.5441
Where 𝑘 = 𝑦 − ℎ𝑖
22. PREDICTIVE MODEL USING DIFFERENCE BETWEEN HEALTHY
AND INJURY
𝑡 𝑦,𝑖 =
log(
𝑘
𝑐𝑖 − ℎ𝑖
−0.5441
Optimal k:
0.0014
Mean:
6.97 Days
Standard Deviation:
1.52 Days
25. PREDICTIVE MODEL USING RATIOS
With β selected as 0.5696
𝑟2 is optimized at 0.43
𝑦𝑖
ℎ𝑖
= (
𝑐𝑖
ℎ𝑖
− 1 )𝑒−0.5696𝑡 + 1
26. PREDICTIVE MODEL USING RATIOS
To predict days until RTP:
𝑡 𝑦,𝑖 =
𝑙𝑜𝑔(
𝑝∗ℎ 𝑖
𝑐 𝑖−ℎ 𝑖
)
−0.5441
Where
𝑝 =
𝑦𝑖
ℎ𝑖
− 1
27. PREDICTIVE MODEL USING RATIOS OF
HEALTHY TO INJURY
Optimal p:
0.008
Mean:
6.97 Days
Standard Deviation:
1.50 Days
𝑡 𝑦,𝑖 =
𝑙𝑜𝑔(
𝑝∗ℎ 𝑖
𝑐 𝑖−ℎ 𝑖
)
−0.5441
30. SYMPTOM LINEAR MODELING:
SINGLE DESCRIPTIVE VARIABLE
Where:
y: Represents the response variable, days since injury
x: Represents the predictor variable, a symptom
𝑏0: Is the value of y when x is zero
𝑏1: The amount y changes when x increases by 1
𝑦 = 𝑏0 + 𝑏1 𝑥
31. SYMPTOM LINEAR MODELING:
MULTIPLE DESCRIPTIVE VARIABLES
Where:
n: Is the number of predictor variables used in the model
𝑥𝑙: Is a predictor variable value, where 𝑙 = 1, 2, … , 𝑛 are
symptoms
𝑏𝑙: The amount y changes when 𝑥𝑙 increases by 1
𝑦 = 𝑏0 + 𝑏1 𝑥1 + 𝑏2 𝑥2+. . . +𝑏 𝑛 𝑥 𝑛
38. CONCLUSIONS:
DETERMINING A CONCUSSION OFF OF STOP WATCH
TESTING
The reaction times taken after injury were significantly different from
those of healthy times taken at the beginning of the season.
39. CONCLUSIONS:
RETURN TO PLAY
𝑡 𝑦,𝑖 =
𝑙𝑜𝑔(
0.008099
𝑐𝑖 − ℎ𝑖
−0.5441
Best Calculator:
Roughly 97.5% of cases recover in 10 Days.
40. CONCLUSIONS:
CONCUSSION SYMPTOMS
Headache, dizziness, fatigue, heightened feelings, and feeling nauseas
showed significance when looking at reaction time.
Headache showed significance when looking at days since injury.