predicting inpatient census with viral surveillence data · learning objectives •identify...
TRANSCRIPT
![Page 1: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/1.jpg)
Predicting Inpatient Census with Viral Surveillence Data
Nephi Walton, MD, MSc; Pediatrics/Genetics Fellow; Washington
University School of Medicine
Gary D. Miner, PhD; Healthcare innovation Lead; Dell Inc
March 3, 2016
![Page 2: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/2.jpg)
Dr. Gary Miner • Works for Dell Statistica - Software as Healthcare Innovation Lead • Doctorate from the University of Kansas, Post-doctoral fellowships in
Behavioral Genetics (University of Minnesota) and Psychiatric Epidemiology and Biostatistics (NIH/University of Iowa)
• Research focuses on: Familial Alzheimer’s Disease • Organized the first International Conference on Familial Alzheimer’s
Disease held in Tulsa, OK. • Prolific writer with numerous works to his name including:
Familial Alzheimer’s Disease, Caring for Alzheimer’s Patients, and Practical Predictive Analytics & Decisioning Systems for Medicine.
Dr. Nephi Walton • Works at Washington University School of Medicine – Pediatrics/Genetics Fellow • Masters from University of Utah School of Medicine Department of Biomedical
Informatics • MD from the University of Utah School of Medicine • Research focuses on: Predictive Analytics and Genomics • Author of several papers and books on predictive analytics and machine learning
in medicine.
![Page 3: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/3.jpg)
Conflict of Interest
Gary Miner, PhD employed by Dell
He is also an instructor at the University of California
Has no real or apparent conflicts of interest to report.
Nephi Walton, MD, MSc employed by Washington University School of Medicine. He is also an instructor at the University of California Has no real or apparent conflicts of interest to report.
![Page 4: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/4.jpg)
Agenda
• Short General Introduction of “THE PROBLEM” - Gary Miner
• SOLUTION(s) using specific examples - Nephi
Walton
![Page 5: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/5.jpg)
Learning Objectives
• Identify opportunities for using predictive analytics and big data sets to improve resource allocation
• Analyze the practical steps needed to take advantage of the opportunities
• Identify key challenges in using analytics to predict inpatient census
• Demonstrate a practical study of using predictive analytics to help with planning, guide staffing and resource allocation and reduce the cost of care
![Page 6: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/6.jpg)
An Introduction of How
Benefits Were Realized for
the Value of Health IT
SATISFACTION: Hospital staff are happier with more
predictable schedules. Patient satisfaction improved by
better care with an appropriate nurse-to-patient ratio.
TREATMENT: Treatment is improved by having the
appropriate resources available to meet patient needs.
ELECTRONIC INFORMATION: By harnessing the
power of data and predictive analytics, we take the
guess work out patient care and resource allocation by
making accurate predictions about inpatient census.
PREVENTATION & PATIENT EDUCATION:
Inadequate staffing and insufficient resources can lead
to adverse events in a clinical setting. These models
help prevent adverse events by ensuring appropriate
nurse-to-patient ratios and adequate resources to
handle patient workloads.
SAVINGS: Costs are reduced from avoiding adverse
events and better resource prediction and allocation
which reduces unnecessary expenses such as
overstaffing.
http://www.himss.org/ValueSuite
![Page 7: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/7.jpg)
GARY MINER - Introduction:
This challenge is not a new thing ……. Even in 2002 this was being discussed, as per this New England Journal of Medicine paper: http://www.nejm.org/doi/full/10.1056/NEJMsa012247?viewType=Print&viewClass=Print&
SPECIAL ARTICLE
Nurse-Staffing Levels and the Quality of Care in Hospitals Jack Needleman, Ph.D., Peter Buerhaus, Ph.D., R.N., Soeren Mattke, M.D., M.P.H., Maureen Stewart, B.A., and Katya Zelevinsky; N Engl J Med 2002; 346:1715-1722May 30, 2002DOI: 10.1056/NEJMsa012247
THE HEALTHCARE CHALLENGE: Predict patient census at a tertiary care children’s hospital to help with resource allocation, preventative measures, and staffing.
![Page 8: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/8.jpg)
2010
Medical epidemics and pandemics in children … From: MICHIGAN DEPARTMENT OF HEALTH AND HUMAN SERVICES, 2010; Medical surge and Medical Surge Triggers…
https://www.michigan.gov/documents/mdch/Planning_for_Children_in_Di
sasters_15_495237_7.pdf
![Page 9: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/9.jpg)
Another example is the NATIONAL ACDEMY OF SCIENCES stressing things like: “Hospitals and acute care facilities providing acute medical care to the community have a “duty to plan” for mass casualty incidents, including planning for the expansion of clinical operations, commonly referred to as
…… surge capacity…”
From their report:
Crisis Standards of Care: A Systems Framework for Catastrophic Disaster Response; Institute of Medicine. Washington (DC): 2012 Mar 21.
2012
![Page 10: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/10.jpg)
It all started with a different kind of mining……
![Page 11: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/11.jpg)
Predict patient census at a tertiary care children’s hospital to help with resource allocation, preventative measures,
and staffing.
The Healthcare Challenge
![Page 12: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/12.jpg)
• Assessed the top reasons for patient admissions.
• The largest by far was viral respiratory illness especially during time of peak census.
• Admissions were further broken down by virus and
Respiratory Syncytial Virus (RSV) was found to be the highest cause of respiratory viral admissions.
Step 1 – Defining the Problem
![Page 13: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/13.jpg)
Goal: Predict when RSV outbreaks will occur.
![Page 14: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/14.jpg)
Step 2 – Selecting Variables
![Page 15: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/15.jpg)
Findings
• We were able to forecast outbreaks up to 3 weeks in advance with sensitivity of 67% and specificity from 94% to 100%.
• Weather variable’s related to RSV are similar to those that cause inversions in the Salt Lake Valley.
• Interesting findings in relation to PM2.5, more data needed.
![Page 16: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/16.jpg)
Next Challenge
To predict weekly average census for hospital staffing.
![Page 17: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/17.jpg)
Background – Nurse Staffing
• Appropriate nurse staffing has been associated with less hospital-related mortality and adverse events.
• Methods have been proposed to handle nurse staffing.
• It has been suggested that the most difficult part of staffing is
predicting inpatient census.
![Page 18: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/18.jpg)
• Capable of modeling complex non-linear relationships.
• Modeled after the workings of neurons in the human brain.
• Model is a black box that cannot easily be interpreted.
Background – Neural Networks
![Page 19: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/19.jpg)
Findings and Gaps
• Emergency Department Admissions prediction models found good results based calendar variability, meteorological data was not effective.
• Minimal literature on forecasting models for patient census. Models that do exist are Auto-Regressive Integrated Moving Average (ARIMA) time series models based on past census numbers.
• Only data mining based patient census forecasting model found was authors previous study using neural network time series models.
![Page 20: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/20.jpg)
Goal To Develop a real-time model to forecast patient census in a
tertiary care children’s hospital.
![Page 21: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/21.jpg)
Hypothesis
• Calendar variables, viral data, and behavioral inputs can be used with machine learning techniques to forecast patient census at a tertiary care children’s hospital.
![Page 22: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/22.jpg)
Methods – Assessment of factors affecting patient census Top Sources of Admissions • Bronchiolitis • Chemotherapy • Dehydration/Gastroenteritis • Asthma • Seizure / Febrile Convulsions
![Page 23: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/23.jpg)
Methods – Assessment of factors affecting patient census. `
• Calendar • Month • Day of Week
• Behavioral
• School • Holidays • BYU/Utah Games
r
![Page 24: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/24.jpg)
Methods – Data Sources
• Lab Data – Viral Tests - Intermountain Health Care (IHC) • Weekly average • 7 day slope • 14 day slope
• Inpatient Elective Surgery Census – IHC • Weekly average
• Inpatient Census Primary Children’s Medical Center (PCMC) • Weekly average
• Federal Holidays - U.S. Office of Personnel Management
• BYU-Utah Games – “Holy War” - Wikipedia
• Data captured from 2007 - 2011
![Page 25: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/25.jpg)
• Clinical Data obtained for Salt Lake County from Intermountain Healthcare Hospitals and Clinics.
• Intermountain Healthcare is the largest healthcare provider in Utah and includes Primary Children’s Medical Center.
Methods – Geographic Region
![Page 26: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/26.jpg)
Methods – Variable Selection
• Feature selection was performed on a continuous dependent variable and all variables with a p-value < .01 were selected for inclusion in the model.
• Feature selection was performed separately for each week in
advance the prediction was made up to 3 weeks, resulting in 3 sets of variables.
![Page 27: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/27.jpg)
Methods – ANN Prediction Model
• Weekly patient census averages calculated based on average patient census from Tues through Sunday.
• Input variables selected by feature selection.
• Output variable is continuous variable representing average weekly census of week predicted.
• Years 2007 – 2010 used for training (85%) and testing (15%), 2011-2012 season used for validation.
• Ran 100 ANN models using Statistica Data Miner and selected best one and applied it to the validation set.
![Page 28: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/28.jpg)
Results – Feature Selection
• Holidays • Month • School in Session • Current Weekly Average Census • Total Respiratory Viral Tests • Adenovirus • Human Metapneumovirus • Influenza A • Influenza B • Rhino Virus • RSV
![Page 29: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/29.jpg)
Results – ANN
• 1 Week in advance correlation coefficient 91%
• 2 Weeks in advance correlation coefficient 80% • 3 Weeks in advance correlation coefficient 88%
![Page 30: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/30.jpg)
Results – ANN
• Using viral test results outside of PCMC viral activity decreased performance in the “1 week in advance” model, but increased performance when predicting further out.
• Removing actual elective surgery number inputs from the model did not affect performance in any of the models.
![Page 31: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/31.jpg)
80
90
100
110
120
130
140
150
160
170
180
190
200
210
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
Ho
sp
ita
l C
en
su
s
Week of the Year
Actual
Predicted
Artificial Neural Network Predictions 1 Week in Advance
![Page 32: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/32.jpg)
80
90
100
110
120
130
140
150
160
170
180
190
200
210
220
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Ho
sp
ita
l C
en
su
s
Week of the Year
Actual
Predicted
Artificial Neural Network Predictions 2 Weeks in Advance
![Page 33: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/33.jpg)
80
90
100
110
120
130
140
150
160
170
180
190
200
210
220
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36
Ho
sp
ita
l C
en
su
s
Week of the Year
Actual
Predicted
Artificial Neural Network Predictions 3 Weeks in Advance
![Page 34: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/34.jpg)
Comparison to ARIMA models
• Performance was far better but not directly comparable to the ARIMA models used.
• ANN correlation for 1, 2, and 3 week forecasting are 0.91,
0.80, and 0.88, which is higher than the best ARIMA value of 0.71 on the same data.
![Page 35: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/35.jpg)
Discussion – Confounders
97-98 98-99 99-00 00-01 01-02 02-03 03-04 04-05 05-06 06-07 07-08 08-09 09-10 10-110
10
20
30
40
50
60
70
80
90
100
Season
av
era
ge
LO
S h
ou
rs
includes RivertonPediatric unit
SOURCE: EDW CaseMix & Transactions
Average Length of Stay Hours, PCMC Bronchiolitis AdmissionsSOI 1 or 2, exclusions applied
69.40
509
64.66
369
59.43
523
59.04
430
56.18
649
59.58
503
59.02
641
67.45
385
76.48
625
75.94
588
69.85
911
67.57
804
63.27
725
48.62
837Average
Cases
• H1N1 outbreak of 2009.
• Changes in RTU have lessened patient stay for 2010-2011 season.
![Page 36: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/36.jpg)
Discussion
• Elective surgery does not have a significant impact on census surges.
• Outside viral activity is better for predicting long term trends.
![Page 37: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/37.jpg)
Conclusions
• ANN models can be effective for forecasting patient census up to 3 weeks in advance.
• Results are comparable to if not better than reported performance of ARIMA models.
• Respiratory viral data, calendar information, holidays, and school attendance appear to be good predictive variables.
• Elective surgeries have minimal impact on census prediction.
![Page 38: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/38.jpg)
Future directions
• Modify model to work on a more granular level.
• Build Disease specific models i.e. asthma, gastroenteritis.
• Deploy model and monitor prospectively. • Include other measures of community disease burden, such as
ICD9 coded visits from urgent cares, in the model to improve performance.
![Page 39: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/39.jpg)
Back to the Mining Analogy
Harnessing the signals from our own body and using our genetic information (“make”/”model”) we may be able to predict failure/disease and intervene before failure occurs.
![Page 40: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/40.jpg)
An Introduction of How
Benefits Were Realized for
the Value of Health IT
SATISFACTION: Hospital staff are happier with more
predictable schedules. Patient satisfaction improved by
better care with an appropriate nurse-to-patient ratio.
TREATMENT: Treatment is improved by having the
appropriate resources available to meet patient needs.
ELECTRONIC INFORMATION: By harnessing the
power of data and predictive analytics, we take the
guess work out patient care and resource allocation by
making accurate predictions about inpatient census.
PREVENTATION & PATIENT EDUCATION:
Inadequate staffing and insufficient resources can lead
to adverse events in a clinical setting. These models
help prevent adverse events by ensuring appropriate
nurse-to-patient ratios and adequate resources to
handle patient workloads.
SAVINGS: Costs are reduced from avoiding adverse
events and better resource prediction and allocation
which reduces unnecessary expenses such as
overstaffing.
http://www.himss.org/ValueSuite
![Page 41: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/41.jpg)
Acknowledgements
• Per Gesteland MD, MS • Seth Andrews • Kent Korgenski • Darren Johnson • Julio Facelli PhD • Intermountain Healthcare - Primary Children’s Hospital • Washington University - St Louis Children’s Hospital • University of Utah Department of Biomedical Informatics
![Page 42: Predicting Inpatient Census with Viral Surveillence Data · Learning Objectives •Identify opportunities for using predictive analytics and big data sets to improve resource allocation](https://reader036.vdocument.in/reader036/viewer/2022070804/5f035b0e7e708231d408ce95/html5/thumbnails/42.jpg)
Nephi Walton MD, MS Washington University, St Louis MO
@geneticks
Gary D. Miner PhD
Dell, Dell Software, Tulsa OK @Gary_Miner
Questions ???