SlideShare a Scribd company logo
1 of 1
Download to read offline
© Copyright 2014 Gray Matter Analytics. All rights reserved.
Clustering Medical Data to Predict the Likelihood of Diseases
Objective: Build a statistical strategy from electronic medical data that can identify the population of a
homogenous set of complex patients who may benefit from targeted care management strategies.
Methodology: We identified 934 patients, some who had a medical history of various conditions and some
who are currently undergoing medical treatment for a specific condition. We used an agglomerative
hierarchical clustering method to identify clinically relevant subgroups with similar conditions. Clustering
compared each member based on data collected for variables such as - (i) Most recent drug fill, (ii) disease
conditions over the past 2 years, (iii) Diagnosis group code, (iv) Emergency room admission count, and (v)
Immunizations over the past 4 years. Patients were then added to a different cluster based on a
comparison of their medical data similarities. The results enabled us to show a clustering model of patients
who were at high risk of acquiring a specific chronic disease and expected to undergo the same kind of
treatment similar to other patients in the same cluster with the same chronic disease. With this method,
medical treatment administered prior to a diagnosis could possibly avoid the risk of acquiring those chronic
diseases.
For example: Let’s say patient “A” was diagnosed with diabetes. Prior to the diagnosis, they were treated
for ailments such as fatigue and blurred vision. Let’s say Patient “B” has been treated for fatigue and is
currently being treated for blurred vision. Because of the similarities in symptoms between the two
patients, there is a strong possibility Patient “B” may be diagnosed with diabetes as well. In this case,
clustering compares each patient’s clinical activities, and, based on similarity, puts patients into a different
cluster.
Benefit: Enables prediction of a diseased condition for a specific patient based on comparison between
other patients who went through a similar pattern of symptoms or treatment before they were diagnosed
with the same type of disease.
Cluster analysis can be used to address a wide variety of important issues for individual and the population
level of healthcare such as - (i) For any given cluster, one might track the outcome of patients going through
different clinical treatments and be able to identify which treatment is most effective or has the least side
effects. (ii) Clustering patients allows adoption of a potentially changing medical landscape.
For this example, if a new disease appears with a particular symptom, the model will identify the disease by
the symptom and create a new cluster of patients who have that disease.
Working with Gray Matter Analytics
Gray Matter Analytics provides professional services that empower our clients in healthcare and financial
services for success, enabling you to use data analytics and predictive modeling to gain a distinct edge over
the competition. We help you gain insight into data you need in order to make informed decisions and
ensure you have the technology platform to enable a data-driven environment.
visit www.graymatteranalytics.com.

More Related Content

What's hot

Comorbidities Present in the Alopecia Areata Registry, Biobank & Clinical Tri...
Comorbidities Present in the Alopecia Areata Registry, Biobank & Clinical Tri...Comorbidities Present in the Alopecia Areata Registry, Biobank & Clinical Tri...
Comorbidities Present in the Alopecia Areata Registry, Biobank & Clinical Tri...National Alopecia Areata Foundation
 
Finding patterns of chronic disease and medication prescriptions from a large...
Finding patterns of chronic disease and medication prescriptions from a large...Finding patterns of chronic disease and medication prescriptions from a large...
Finding patterns of chronic disease and medication prescriptions from a large...Graph-TA
 
Innovation in patient care
Innovation in patient careInnovation in patient care
Innovation in patient careAmit Garg
 
A Two-sample Approach for State Estimates of a Chronic Condition Outcome
A Two-sample Approach for State Estimates of a Chronic Condition OutcomeA Two-sample Approach for State Estimates of a Chronic Condition Outcome
A Two-sample Approach for State Estimates of a Chronic Condition Outcomesoder145
 
BIOSTATISTICS AND GENITICS
BIOSTATISTICS AND GENITICSBIOSTATISTICS AND GENITICS
BIOSTATISTICS AND GENITICSriancopper
 
A review of use of enantiomers in homeopathy
A review of use of enantiomers in homeopathyA review of use of enantiomers in homeopathy
A review of use of enantiomers in homeopathyhome
 
Pistoia Alliance datathon for drug repurposing for rare diseases
Pistoia Alliance datathon for drug repurposing for rare diseasesPistoia Alliance datathon for drug repurposing for rare diseases
Pistoia Alliance datathon for drug repurposing for rare diseasesPistoia Alliance
 
Resource Guide to Informatics Standards
Resource Guide to Informatics StandardsResource Guide to Informatics Standards
Resource Guide to Informatics Standardsjetweedy
 
A web/mobile decision support system to improve medical diagnosis using a com...
A web/mobile decision support system to improve medical diagnosis using a com...A web/mobile decision support system to improve medical diagnosis using a com...
A web/mobile decision support system to improve medical diagnosis using a com...TELKOMNIKA JOURNAL
 
Journal review of EHR use and benefits
Journal review of EHR use and benefitsJournal review of EHR use and benefits
Journal review of EHR use and benefitsJ. Don Soriano
 
2010 03 18 Mod 4 Class 2
2010 03 18 Mod 4 Class 22010 03 18 Mod 4 Class 2
2010 03 18 Mod 4 Class 2growell
 
Future of Healthcare Forum (Digital Health 2017) - Andrew Satz
Future of Healthcare Forum (Digital Health 2017) - Andrew SatzFuture of Healthcare Forum (Digital Health 2017) - Andrew Satz
Future of Healthcare Forum (Digital Health 2017) - Andrew SatzOZ Digital Consulting
 
Federal HAI Data Summit May 2012 plenary two-master_slides noel slides 11 t...
Federal HAI Data Summit May 2012   plenary two-master_slides noel slides 11 t...Federal HAI Data Summit May 2012   plenary two-master_slides noel slides 11 t...
Federal HAI Data Summit May 2012 plenary two-master_slides noel slides 11 t...Noel Eldridge
 
Overall patient satisfaction was significantly higher in homeopathic than in ...
Overall patient satisfaction was significantly higher in homeopathic than in ...Overall patient satisfaction was significantly higher in homeopathic than in ...
Overall patient satisfaction was significantly higher in homeopathic than in ...home
 

What's hot (20)

Towards a learning health system
Towards a learning health systemTowards a learning health system
Towards a learning health system
 
Comorbidities Present in the Alopecia Areata Registry, Biobank & Clinical Tri...
Comorbidities Present in the Alopecia Areata Registry, Biobank & Clinical Tri...Comorbidities Present in the Alopecia Areata Registry, Biobank & Clinical Tri...
Comorbidities Present in the Alopecia Areata Registry, Biobank & Clinical Tri...
 
RDD Conf Day 2: Josh Lounsberry (Canadian Neuromuscular Disease Network)
RDD Conf Day 2: Josh Lounsberry (Canadian Neuromuscular Disease Network)RDD Conf Day 2: Josh Lounsberry (Canadian Neuromuscular Disease Network)
RDD Conf Day 2: Josh Lounsberry (Canadian Neuromuscular Disease Network)
 
Finding patterns of chronic disease and medication prescriptions from a large...
Finding patterns of chronic disease and medication prescriptions from a large...Finding patterns of chronic disease and medication prescriptions from a large...
Finding patterns of chronic disease and medication prescriptions from a large...
 
Innovation in patient care
Innovation in patient careInnovation in patient care
Innovation in patient care
 
A Two-sample Approach for State Estimates of a Chronic Condition Outcome
A Two-sample Approach for State Estimates of a Chronic Condition OutcomeA Two-sample Approach for State Estimates of a Chronic Condition Outcome
A Two-sample Approach for State Estimates of a Chronic Condition Outcome
 
BIOSTATISTICS AND GENITICS
BIOSTATISTICS AND GENITICSBIOSTATISTICS AND GENITICS
BIOSTATISTICS AND GENITICS
 
Week 12 LVT
Week 12 LVTWeek 12 LVT
Week 12 LVT
 
A review of use of enantiomers in homeopathy
A review of use of enantiomers in homeopathyA review of use of enantiomers in homeopathy
A review of use of enantiomers in homeopathy
 
NETWORK OF DISEASES AND ITS ENDOWMENT TOWARDS DISEASE
NETWORK OF DISEASES AND ITS ENDOWMENT TOWARDS DISEASE NETWORK OF DISEASES AND ITS ENDOWMENT TOWARDS DISEASE
NETWORK OF DISEASES AND ITS ENDOWMENT TOWARDS DISEASE
 
Pistoia Alliance datathon for drug repurposing for rare diseases
Pistoia Alliance datathon for drug repurposing for rare diseasesPistoia Alliance datathon for drug repurposing for rare diseases
Pistoia Alliance datathon for drug repurposing for rare diseases
 
Resume
ResumeResume
Resume
 
Resource Guide to Informatics Standards
Resource Guide to Informatics StandardsResource Guide to Informatics Standards
Resource Guide to Informatics Standards
 
A web/mobile decision support system to improve medical diagnosis using a com...
A web/mobile decision support system to improve medical diagnosis using a com...A web/mobile decision support system to improve medical diagnosis using a com...
A web/mobile decision support system to improve medical diagnosis using a com...
 
Journal review of EHR use and benefits
Journal review of EHR use and benefitsJournal review of EHR use and benefits
Journal review of EHR use and benefits
 
Trust, Respect, and Reciprocity
Trust, Respect, and ReciprocityTrust, Respect, and Reciprocity
Trust, Respect, and Reciprocity
 
2010 03 18 Mod 4 Class 2
2010 03 18 Mod 4 Class 22010 03 18 Mod 4 Class 2
2010 03 18 Mod 4 Class 2
 
Future of Healthcare Forum (Digital Health 2017) - Andrew Satz
Future of Healthcare Forum (Digital Health 2017) - Andrew SatzFuture of Healthcare Forum (Digital Health 2017) - Andrew Satz
Future of Healthcare Forum (Digital Health 2017) - Andrew Satz
 
Federal HAI Data Summit May 2012 plenary two-master_slides noel slides 11 t...
Federal HAI Data Summit May 2012   plenary two-master_slides noel slides 11 t...Federal HAI Data Summit May 2012   plenary two-master_slides noel slides 11 t...
Federal HAI Data Summit May 2012 plenary two-master_slides noel slides 11 t...
 
Overall patient satisfaction was significantly higher in homeopathic than in ...
Overall patient satisfaction was significantly higher in homeopathic than in ...Overall patient satisfaction was significantly higher in homeopathic than in ...
Overall patient satisfaction was significantly higher in homeopathic than in ...
 

Similar to Clustering Medical Data to Predict the Likelihood of Diseases V2 1

Prospective Payment System
Prospective Payment SystemProspective Payment System
Prospective Payment SystemKaty Allen
 
Classifying Readmissions of Diabetic Patient Encounters
Classifying Readmissions of Diabetic Patient EncountersClassifying Readmissions of Diabetic Patient Encounters
Classifying Readmissions of Diabetic Patient EncountersMayur Srinivasan
 
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RI...
 PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RI... PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RI...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RI...hiij
 
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...hiij
 
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...hiij
 
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...hiij
 
Comparing Patients’ Experiences in Three Differentiated Service Delivery Mode...
Comparing Patients’ Experiences in Three Differentiated Service Delivery Mode...Comparing Patients’ Experiences in Three Differentiated Service Delivery Mode...
Comparing Patients’ Experiences in Three Differentiated Service Delivery Mode...Ferdinand C Mukumbang
 
Enabling clinical trial expansion
Enabling clinical trial expansionEnabling clinical trial expansion
Enabling clinical trial expansionIMSHealthRWES
 
Population Health Management
Population Health ManagementPopulation Health Management
Population Health ManagementVitreosHealth
 
Clinical Data Science and its Future
Clinical Data Science and its FutureClinical Data Science and its Future
Clinical Data Science and its FutureEditorIJTSRD1
 
Machine Learning applied to heart failure readmissions
Machine Learning applied to heart failure readmissionsMachine Learning applied to heart failure readmissions
Machine Learning applied to heart failure readmissionsJohn Frias Morales, DrBA, MS
 
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...hiij
 
Real world Evidence and Precision medicine bridging the gap
Real world Evidence and Precision medicine bridging the gapReal world Evidence and Precision medicine bridging the gap
Real world Evidence and Precision medicine bridging the gapClinosolIndia
 
PARR-combined-predictive-model-final-report-dec06
PARR-combined-predictive-model-final-report-dec06PARR-combined-predictive-model-final-report-dec06
PARR-combined-predictive-model-final-report-dec06Nadya Filipova
 
Personalized Medicine and Predictive Analytics A Review of Computational Methods
Personalized Medicine and Predictive Analytics A Review of Computational MethodsPersonalized Medicine and Predictive Analytics A Review of Computational Methods
Personalized Medicine and Predictive Analytics A Review of Computational Methodsijtsrd
 
Machine learning and operations research to find diabetics at risk for readmi...
Machine learning and operations research to find diabetics at risk for readmi...Machine learning and operations research to find diabetics at risk for readmi...
Machine learning and operations research to find diabetics at risk for readmi...John Frias Morales, DrBA, MS
 
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...hiij
 
The Role of Real-World Data in Clinical Development
The Role of Real-World Data in Clinical DevelopmentThe Role of Real-World Data in Clinical Development
The Role of Real-World Data in Clinical DevelopmentCovance
 

Similar to Clustering Medical Data to Predict the Likelihood of Diseases V2 1 (20)

Prospective Payment System
Prospective Payment SystemProspective Payment System
Prospective Payment System
 
Classifying Readmissions of Diabetic Patient Encounters
Classifying Readmissions of Diabetic Patient EncountersClassifying Readmissions of Diabetic Patient Encounters
Classifying Readmissions of Diabetic Patient Encounters
 
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RI...
 PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RI... PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RI...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RI...
 
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
 
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
 
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
PERSONALIZED MEDICINE SUPPORT SYSTEM: RESOLVING CONFLICT IN ALLOCATION TO RIS...
 
Comparing Patients’ Experiences in Three Differentiated Service Delivery Mode...
Comparing Patients’ Experiences in Three Differentiated Service Delivery Mode...Comparing Patients’ Experiences in Three Differentiated Service Delivery Mode...
Comparing Patients’ Experiences in Three Differentiated Service Delivery Mode...
 
CPT.BigHealthcareData.2016
CPT.BigHealthcareData.2016CPT.BigHealthcareData.2016
CPT.BigHealthcareData.2016
 
Enabling clinical trial expansion
Enabling clinical trial expansionEnabling clinical trial expansion
Enabling clinical trial expansion
 
Population Health Management
Population Health ManagementPopulation Health Management
Population Health Management
 
Clinical Data Science and its Future
Clinical Data Science and its FutureClinical Data Science and its Future
Clinical Data Science and its Future
 
Machine Learning applied to heart failure readmissions
Machine Learning applied to heart failure readmissionsMachine Learning applied to heart failure readmissions
Machine Learning applied to heart failure readmissions
 
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
 
Real world Evidence and Precision medicine bridging the gap
Real world Evidence and Precision medicine bridging the gapReal world Evidence and Precision medicine bridging the gap
Real world Evidence and Precision medicine bridging the gap
 
PARR-combined-predictive-model-final-report-dec06
PARR-combined-predictive-model-final-report-dec06PARR-combined-predictive-model-final-report-dec06
PARR-combined-predictive-model-final-report-dec06
 
Personalized Medicine and Predictive Analytics A Review of Computational Methods
Personalized Medicine and Predictive Analytics A Review of Computational MethodsPersonalized Medicine and Predictive Analytics A Review of Computational Methods
Personalized Medicine and Predictive Analytics A Review of Computational Methods
 
Comorbilidades
ComorbilidadesComorbilidades
Comorbilidades
 
Machine learning and operations research to find diabetics at risk for readmi...
Machine learning and operations research to find diabetics at risk for readmi...Machine learning and operations research to find diabetics at risk for readmi...
Machine learning and operations research to find diabetics at risk for readmi...
 
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
A KNOWLEDGE DISCOVERY APPROACH FOR BREAST CANCER MANAGEMENT IN THE KINGDOM OF...
 
The Role of Real-World Data in Clinical Development
The Role of Real-World Data in Clinical DevelopmentThe Role of Real-World Data in Clinical Development
The Role of Real-World Data in Clinical Development
 

Clustering Medical Data to Predict the Likelihood of Diseases V2 1

  • 1. © Copyright 2014 Gray Matter Analytics. All rights reserved. Clustering Medical Data to Predict the Likelihood of Diseases Objective: Build a statistical strategy from electronic medical data that can identify the population of a homogenous set of complex patients who may benefit from targeted care management strategies. Methodology: We identified 934 patients, some who had a medical history of various conditions and some who are currently undergoing medical treatment for a specific condition. We used an agglomerative hierarchical clustering method to identify clinically relevant subgroups with similar conditions. Clustering compared each member based on data collected for variables such as - (i) Most recent drug fill, (ii) disease conditions over the past 2 years, (iii) Diagnosis group code, (iv) Emergency room admission count, and (v) Immunizations over the past 4 years. Patients were then added to a different cluster based on a comparison of their medical data similarities. The results enabled us to show a clustering model of patients who were at high risk of acquiring a specific chronic disease and expected to undergo the same kind of treatment similar to other patients in the same cluster with the same chronic disease. With this method, medical treatment administered prior to a diagnosis could possibly avoid the risk of acquiring those chronic diseases. For example: Let’s say patient “A” was diagnosed with diabetes. Prior to the diagnosis, they were treated for ailments such as fatigue and blurred vision. Let’s say Patient “B” has been treated for fatigue and is currently being treated for blurred vision. Because of the similarities in symptoms between the two patients, there is a strong possibility Patient “B” may be diagnosed with diabetes as well. In this case, clustering compares each patient’s clinical activities, and, based on similarity, puts patients into a different cluster. Benefit: Enables prediction of a diseased condition for a specific patient based on comparison between other patients who went through a similar pattern of symptoms or treatment before they were diagnosed with the same type of disease. Cluster analysis can be used to address a wide variety of important issues for individual and the population level of healthcare such as - (i) For any given cluster, one might track the outcome of patients going through different clinical treatments and be able to identify which treatment is most effective or has the least side effects. (ii) Clustering patients allows adoption of a potentially changing medical landscape. For this example, if a new disease appears with a particular symptom, the model will identify the disease by the symptom and create a new cluster of patients who have that disease. Working with Gray Matter Analytics Gray Matter Analytics provides professional services that empower our clients in healthcare and financial services for success, enabling you to use data analytics and predictive modeling to gain a distinct edge over the competition. We help you gain insight into data you need in order to make informed decisions and ensure you have the technology platform to enable a data-driven environment. visit www.graymatteranalytics.com.