@suchisaria - cra · may 13th, 1999 — april 1, 2012 example: septicemia is the 11th leading cause...
TRANSCRIPT
Making ‘Meaningful Use’ more meaningful
Suchi Saria Assistant Professor of Computer Science, Statistics, Biostatistics, and Health Policy
Malone Center for Engineering in Healthcare,
Johns Hopkins University
@suchisaria
Discrete Events:
Laboratory
Interventions: Medicines, Procedures
Continuous p
hysiologic
measurements
Progress notes
Imaging Data
Administrative Claims
Genomic data
Sensors & Devices
Electronic Health Data
From 2008 to 2014, hospitals with EHRs rose to 75% from 9%, and in doctors’ offices rose to 51% from 17%.
$32+ billion effort went into digitization. Benefits?
May 13th, 1999 — April 1, 2012
Example: Septicemia is the 11th leading cause of death in the US
3.9 times higher mortality rate 2.4 times longer mean Length of Stay (LOS)
2.7 times higher mean cost R. L. Fuller, et al., Health Care Financing Review, vol. 30, pp. 17-32, 2009
Image credit: The Rory Staunton Foundation
Mortality and length of stay decreased with timely treatment [Kumar et al. 2011]. For every hour that antibiotic treatments were delayed, risk of mortality went up by 7-8%
Can we detect those at risk of this adverse event early?
Early Warning Systems for Potentially Preventable Conditions?
K. Henry, D. Hager, P. Pronovost, S. Saria. A Targeted Real-time Early Warning Score (TREWScore) for Septic Shock. Science Translational Medicine 2015. Vol. 7, Issue 299.
K. Dyagilev, S. Saria. Learning (Predictive) Risk Scores in the Presence of Censoring due to Interventions. Machine Learning, 2015.
Systolic BP
MAP
Heart rate
Creatinine
Hemoglobin
Sodium
Hematocrit
Platelets
INR
PTT
Magnesium
Resp rate
Calcium
BUN
Bicarbonate
Potassium
Hours since hospital admissionK. Henry, D. Hager, P. Pronovost, S. Saria. A Targeted Real-time Early Warning Score (TREWScore) for Septic Shock. Science Translational Medicine 2015. Vol. 7, Issue 299.
K. Dyagilev, S. Saria. Learning (Predictive) Risk Scores in the Presence of Censoring due to Interventions. Machine Learning, 2015.
Seve
rity
Scor
e
Surprising finding: Using routinely collected data, the majority of patients were identified as high risk a median time of 25 hours prior to shock onset.
Mortality and length of stay decreased with timely treatment [Kumar et al. 2011]. For every hour that antibiotic treatments were delayed, risk of mortality went up by 7.6%.
Henry et al., Science Translational Medicine, 2015
Scleroderma - an example
• Systemic autoimmune disease Main symptom: skin fibrosisAffects many visceral organs—lungs, heart, GI tract, kidney, vasculature, and muscles
Affects 300K individuals; 80 other autoimmune diseases — lupus, multiple sclerosis, diabetes, Crohn’s — many of which are systemic & highly multiphenotypic.
Lung
Skin
Kidney
Treatments
●
●
●
●
●●●
●●
●
40
60
80
100
120
0 5 10 15Years Since First Symptom
PFVC
Individualize Prognosis of Disease Activity Trajectory. Why?
●
●
●
●
●●●
●●
●
40
60
80
100
120
0 5 10 15Years Since First Symptom
PFVC
• Will this individual continue to decline? • Should we administer immunosuppressants,
which can be toxic?
Mar
ker f
or lu
ng d
eclin
e (p
FVC
)
●●●
●
●
●
●●● ●
●●
●●●
●●●
●● ●●
● ● ●
●
●
●● ●
●
●●
●
50
70
90
0 5 10 15 20Years Seen
PFVC
● ● ● ●
● ● ●
● ●
●
●●●
●
●●
Low
0
25
50
75
100
125
0 5 10 15 20Years Seen
RP
●
●
●
●
●
●
●●●
●
●●
●
●0
10
20
30
40
0 5 10 15 20Years Seen
TSS
● ● ●
● ● ● ● ● ● ●●●●●
●●
Low
0
25
50
75
100
125
0 5 10 15 20Years Seen
GI
●
●
●
●
●
●
●●
●●
●
●
●●● ●
●
●
●
●
●
●
●
● ●
●
●
●
●
●●●
● ●
●
●
25
50
75
100
0 5 10 15 20Years Seen
PDLC
O● ● ● ● ● ● ● ●●●●●●●
Low
0
25
50
75
100
125
0 5 10 15 20Years Seen
Hear
t
●●●
●
●
●●●● ●
●●● ●
●
●●●
●● ●●
● ●●
●
●
●●●
●
●
●
●
●
50
70
90
0 5 10 15 20Years Seen
PFEV
1
● ● ●
●
●
●
● ● ● ●
●●
●
●
●●
Low
0
25
50
75
100
125
0 5 10 15 20Years Seen
Mus
cle
●●
●
●
0
25
50
75
100
125
0 5 10 15 20Years Seen
RVSP
● ● ● ● ● ● ● ● ●●●●●●●
Low
0
25
50
75
100
125
0 5 10 15 20Years Seen
Kidn
ey
●
●
●
●●
●●
●●●●●
● ●● ● ●
●●
●
50
70
90
0 5 10 15 20Years Seen
PFVC
● ●● ●●●●● ● ●● ●●●●● ●● ●●
Low
0
25
50
75
100
125
0 5 10 15 20Years Seen
RP
●
●● ●●
●
●
● ● ●
● ●●●●● ●● ●●0
10
20
30
40
0 5 10 15 20Years Seen
TSS ●
●● ●
●●●● ● ●● ●●●●● ●● ●●
High
0
25
50
75
100
125
0 5 10 15 20Years Seen
GI
●●
●●●
●●●●● ● ●
● ● ●● ● ●
25
50
75
100
0 5 10 15 20Years Seen
PDLC
O
● ●
●
●●●●● ● ●● ●● ● ●● ●●
Low
0
25
50
75
100
125
0 5 10 15 20Years Seen
Hear
t
●
●
●
●
●●
●●●●
●●
●● ● ● ●
●●
50
70
90
0 5 10 15 20Years Seen
PFEV
1
●● ●●●●● ● ●● ●●●●● ●● ●●
High
0
25
50
75
100
125
0 5 10 15 20Years Seen
Mus
cle
● ● ●●
●
● ●
0
25
50
75
100
125
0 5 10 15 20Years Seen
RVSP
●
●● ●●●●● ● ●● ●●●●● ●● ●
High
0
25
50
75
100
125
0 5 10 15 20Years Seen
Kidn
ey
zi
fi
M
↵
yij tij
Ni
~�g G
~bi
⌃b
~⇢i ~xi⇢
⇡g
B⇢
zi
fi
M
↵
yij tij
Ni
~�g G
~bi
⌃b
~⇢i ~xi⇢
⇡g
B⇢
zi
fi
M
↵
yij tij
Ni
~�g G
~bi
⌃b
~⇢i ~xi⇢
⇡g
B⇢ Schulam, Wigley, Saria, AAAI 2015 Schulam, Saria, NIPS 2015 Schulam, Saria. JMLR 2016
Models for Individualizing Prognoses
Lung
Skin
Vasc
ulat
ure
Schulam, Saria. JMLR 2016
Models for Individualizing Prognoses
13
Physiologic Signals: Collected but Dropped after 48 hours
Heart Rate
Respiratory Rate
Oxygen Saturation
14
Physiologic Signals: Collected but Dropped after 48 hours
Heart Rate
Respiratory Rate
Oxygen Saturation
APGAR (Standard of
care)
CRIB SNAP-II SNAPPE-II Physiscore (Our tool)
Time from birth
5 mins 12 hours 12 hours 12 hours 3 hours after birth
Accuracy 0.69 0.85 0.82 0.87 0.91
Effort Manual Manual Manual Manual Automated on monitor
Invasive testing
15
X X
• Identifies premature infants at risk for major complications • Useful for resource allocation, managing infant transport, staffing-ratio
Physiscore: Automated, Accurate Risk Assessment in Newborns
Science Translational Medicine, 2010
Measurement that is Non-invasive, Low cost, Instrumented on existing device for risk stratification.
Thank [email protected]
www.suchisaria.com@suchisaria
Discussion
• What infrastructure do we need so that we can put these data to use for patients?
• How can we enable more rapid innovation w/ EHRs?