Current Current Environment/ChallengesEnvironment/Challenges Many statistical signals (few with meaning) – Many statistical signals (few with meaning) – easy to get lost in data and signals easy to get lost in data and signals

Many data sources, many data fieldsMany data sources, many data fields Many fields untapped by current statisticsMany fields untapped by current statistics Data sources with preliminary findings often are Data sources with preliminary findings often are

missing final outcome statusmissing final outcome status Lack of data linkage across data sourcesLack of data linkage across data sources

Public health resources are stretched thin with Public health resources are stretched thin with many competing needs – can’t sort through it allmany competing needs – can’t sort through it all

Still - part of the “real story” of disease Still - part of the “real story” of disease burden is buried in these databurden is buried in these data

The Question:The Question:

What approaches What approaches can assist in can assist in

identifying the identifying the actionable actionable

information that information that exists within all of exists within all of these data/signals?these data/signals?

Previous ApproachPrevious Approach Look across “independent” data Look across “independent” data

sources/statistics to corroborate sources/statistics to corroborate signals signals

Problems with this approach:Problems with this approach: Each data source may represent different Each data source may represent different

underlying populationsunderlying populations Difficult to align signals if data sources on Difficult to align signals if data sources on

different time tablesdifferent time tables Signals may not agree and that may not be Signals may not agree and that may not be

good reason to ignore the alertgood reason to ignore the alert

Proposed Proposed Approach:Approach:

Decision Support Decision Support SystemSystem

Decision Support System Decision Support System Incorporated into existing information Incorporated into existing information

systemssystems Organizes data to inform clinical or public Organizes data to inform clinical or public

health responseshealth responses Enables providers to target clinical Enables providers to target clinical

management of medical conditionsmanagement of medical conditions Reminders for vaccines (flu vaccine prompts if Reminders for vaccines (flu vaccine prompts if

> 65 yo) > 65 yo) Screening recommendations (mammograms)Screening recommendations (mammograms) Chronic disease management (asthma, diabetes)Chronic disease management (asthma, diabetes)

Associated with improved outcomesAssociated with improved outcomes Immunizations: Up-to-date ratesImmunizations: Up-to-date rates Decreased emergency room visits for asthmaDecreased emergency room visits for asthma

How Might We Use A Decision How Might We Use A Decision Support System for Support System for

Surveillance?Surveillance? It is not uncommon to have many visits It is not uncommon to have many visits

that fall into a general GI category that fall into a general GI category Able to parse category into more useful Able to parse category into more useful

subcategories:subcategories:GI (common) GI (common) “bloody diarrhea” (moderately “bloody diarrhea” (moderately

common) common) young people with “bloody diarrhea” young people with “bloody diarrhea” (rare)(rare)

With a little extra effort, we now may With a little extra effort, we now may propose propose ShigellaShigella or or E. coliE. coli as a possible as a possible differential diagnosis – thereby framing differential diagnosis – thereby framing decisions for the public health practitionerdecisions for the public health practitioner

Different public health actions are warranted Different public health actions are warranted for a possible for a possible ShigellaShigella or or EE. . colicoli problem vs. problem vs. a general GI flaga general GI flag

Decision Support NeedsDecision Support Needs Prompts for actionPrompts for action

Event detection (Event detection (ALF example to follow)ALF example to follow) Event managementEvent management

Situation awareness, hurricane management, influenza onset, Situation awareness, hurricane management, influenza onset, disaster management, large public events (inauguration), etc. disaster management, large public events (inauguration), etc.

Decision-making may be affected by level of Decision-making may be affected by level of responsible domainresponsible domain FederalFederal StateState LocalLocal

How could this work as integrated electronic How could this work as integrated electronic solution?solution?

Do De Do. . . Skip Do De Do. . . Skip a Fewa Few

Our Usual Way Our Usual Way to Focus on to Focus on

Statistical FlagsStatistical Flags

Rudimentary Example: Rudimentary Example: Alert List Filter (ALF)Alert List Filter (ALF)

Automated way to “digest” many red Automated way to “digest” many red and yellow statistical flagsand yellow statistical flags

Scores each alert based on several Scores each alert based on several criteria (number of visits, “density”, and criteria (number of visits, “density”, and “recency”)“recency”)

Ranks top 20 flags with highest scores Ranks top 20 flags with highest scores and presents them to the user in a tableand presents them to the user in a table

Allows user to visualize the last 7-8 days Allows user to visualize the last 7-8 days of syndromic flags with some inferred of syndromic flags with some inferred importanceimportance

Alert List Filter (ALF) Alert List Filter (ALF) Example (Cont’d)Example (Cont’d)

Instead of focusing solely on the red Instead of focusing solely on the red flag with the greatest number of visits, flag with the greatest number of visits, ALF directed the analyst’s attention to ALF directed the analyst’s attention to an otherwise buried potential cluster an otherwise buried potential cluster of respiratory illness in kidsof respiratory illness in kids

The analyst was then able to visualize The analyst was then able to visualize this “lost” potential cluster by this “lost” potential cluster by querying the system using elements querying the system using elements proposed by the filterproposed by the filter

Decision Support Decision Support ChallengesChallenges

Each level of complexity is a Each level of complexity is a potential barrier to the practitioner’s potential barrier to the practitioner’s ability to find the actionable eventability to find the actionable event

Robust logic must be discovered that Robust logic must be discovered that will propose useful response paths will propose useful response paths thereby limiting complexity for the thereby limiting complexity for the surveillance system usersurveillance system user

Decision Support Decision Support Challenges (Cont’d)Challenges (Cont’d)

In order to accomplish this, work In order to accomplish this, work needs to be done to discover:needs to be done to discover: the the likely events likely events in different data sourcesin different data sources how information can be harvested how information can be harvested from every from every

available elementavailable element of the data of the data how how meaningful differencesmeaningful differences (not just (not just

statistically significant) are detectedstatistically significant) are detected the suggested the suggested responseresponse protocols protocols

This information needs to inform This information needs to inform systems so they can assist systems so they can assist practitioners in quickly making practitioners in quickly making sense of masses of datasense of masses of data