Site Loader
Get an Essay Help
Rock Street 34, San Francisco State

What is the role of CDS
1. Assists providers in accessing relevant knowledge
2. Assist providers in leveraging ELECTRONIC patient data
3. Assists providers in making the RIGHT decision for the RIGHT patient applying the RIGHT knowledge to the RIGHT patient at the RIGHT time
Draw a schematic of a Clinical Decision Support System
What are some components of Clinical Decision Support Systems (CDSS)
1 Event Monitor
2 Data Query Methods
3 Rule builder
4 Rules engine (inference engine)
5 Notification Methods
6 Audit Database for QA, CQI
Describe the components and their purpose
Event Monitor
Triggers a CDS rule for a specific patient if an event occurs (e.g. med order, lab result)
watches for an event to happen
Describe the components and their purpose
Data Query Methods
to retrieve a patient’s data
usually structured data (wt, lab values, etc)
Describe the components and their purpose
Rule builder
Integrated Development Environment (IDE)
Development Wizard – easy to build, limited creativity for providers to make their own rules
Describe the components and their purpose
Rules engine
(inference engine)
Runs CDS rules (e.g. Arden Syntax Medical Logic Module)
Uses Decision Model to decide if Alert needed
Describe the components and their purpose
Notification Methods
Inform provider of event (e.g. error, status change, etc.)
Describe the components and their purpose
Audit Database for QA, CQI
Stores every production run of a CDS rule
what questions do you need to answer to help guide – Key Components of Requirements of CDS Decision Models
1 What is the target patient population? pregnant women
2 When should the algorithm fire? timing, clinical scenario
3 What information is necessary to make a decision?
4 What is the algorithm?
5 What do you want to do with the result of the algorithm? provider notification
Two Types of CDS Systems (CDSS)
1 Non-Patient-Specific CDSS
2 Patient-Specific CDSS
Non-Patient-Specific CDSS
1 Knowledge retrieval systems (KRS) – literature searches pubmed, micromedex, up-to-date
2 Informational Notices (sticky notes, info in ERX records, etc.)
3 Data Mining – Provides population-specific relationships and information (aka Big Data) – reports, treatment guidelines, order sets
True or False
Knowledge retrieval systems (KRS) is clinician initiated
true
Patient-Specific CDSS
1 Triggered by a patient-specific event
2 CDS rule runs a computerized algorithm (Decision Model) to determine if an alert or other notification is needed
Types of Patient-Specific CDSS
1 statistical – quantitative
2 heuristic – qualitative, ALL CDSS, truth tables, boolean logic
Decision Model Examples
1 Boolean Logic (available in all CDS systems)
2 Algorithms and Mathematical Algorithms
3 Statistical Methods
Boolean Logic
1 All CDS systems provide Boolean Logic
Boolean operators: AND, OR and NOT
Boolean rules require at least one criteria
Each criteria will conclude either True or False
CDS rule must concludes TRUE to send an alert
2 Simple CDS rules fire often and cause Alert Fatigue!
Example: Rule with one criteria (e.g. Male)
3 Add more criteria to reduce Alert Fatigue
Example: Male + New Admit + Nurse provider + Open Admission Assessment form -> fire
Linear and logistic Least-square Regression analysis
Detection and prediction of trends
Bayesian statistical models
Examples: estimating PK parameters; classifying ECGs; Diagnosis
Bayesian statistical models favor highest probability answer
Difficult to identify important outliers
True or False
Bayesian statistical models favor lowest probability answer
false
Fuzzy Logic (Fuzzy Arden)
Uncertainty in data or predicted parameters; Fuzzy Logic provides methods for dealing with uncertainty at margins of categories

Crisp Set: crisp boundary
True or False decision
Borderline cases are undetected
Categories with “crisp” BMI ranges for adults:

Fuzzy Set: fuzzy boundary
Gradual transition
Borderline cases are detected
Allows category membership probability (0 1) at borders

HL7 – Arden Syntax Design Philosophy
Allow development of any combination of statistical & heuristic decision models
Discuss CDS Decision Models used to aid decision making and how they relate to clinician responsibility to patients
1 False Positives and False Negatives have unintended consequences
2 Simple CDS rules fire often and cause alert fatigue!
Add more criteria and complex logic to reduce alert fatigue.
3 Bayesian statistical methods favor highest probability answer, making it difficult to identify important outliers
4 Fuzzy Logic provides methods for dealing with uncertainty at margins of categories
Pharmacy Informatics (PI) Landscape
Describe business models and issues
Pharmacy Department vs Information Services/Technology Department
Pharmacy Department
Pros
Pharmacy’s business/values are well understood
Easy access to customers

Cons
Sequestered in pharmacy silo

Information Services/Technology Department
Pros
Easy collaboration
CDS, Nursing, Pharmacy, Radiology, ED, Oncology,

Cons
Removed from their (pharmacy) customers

Pharmacy Informatics (PI) Landscape
Discuss the two types of PI jobs and prerequisites
Pharmacy Analyst: hired and then trained
Informatics Specialist: need credentials before hiring
Pharmacy Analyst
Job Description: Build and maintain pharmacy and drug databases
Prerequisites: Excellent knowledge and experience with medication safety issues
Clinical Pharmacist Informatics Specialist
Job Description: Analytics [informatics] is the discovery and communication of meaningful patterns in data [that] relies on the simultaneous application of statistics, computer programming and operations research to quantify performance.

Prerequisites:Master of Biomedical Informatics, PGY2 residency in Informatics or Clinical Specialty
Quality Improvement Certificate, Statistics especially as it relates to EBM

Key Takeaways
Pharmacy Informatics Landscape
Analyst and Informatics are different jobs
Pharmacy Analyst: hired and then trained
Informatics Specialist: need credentials before hiring
Review slide on PROs and CONs of Business Structure

Related Essay Examples

Post Author: Arnold

Leave a Reply

Your email address will not be published. Required fields are marked *