disability pricing and dental fraud detection: supervised and unsupervised learning october 11, 2007...

Disability PricingDisability Pricingandand

Dental Fraud Detection:Dental Fraud Detection:Supervised and Unsupervised Supervised and Unsupervised

LearningLearningOctober 11, 2007October 11, 2007

Jonathan Polon FSAClaim Analytics Inc.www.claimanalytics.com

• Definitions

• Dental Fraud

• Disability Pricing

AgendaAgenda

DefinitionsDefinitions

Supervised LearningSupervised Learning

• Known outcome associated with each record in training dataset

• Eg, Sports betting: predict winners based on observations from past games

• Objective: build a model to accurately estimate outcomes from predictor variables for each record

• Commonly referred to as predictive modeling

Unsupervised LearningUnsupervised Learning

• No known outcome associated with any record in the training dataset

• Eg, Grocery stores: define types of shoppers and their preferences

• Learning objective: self-organization or clustering; finding structure in the data

Dental Fraud DetectionDental Fraud DetectionUsingUsing

Unsupervised LearningUnsupervised Learning

Dental Fraud Project Dental Fraud Project OverviewOverview• Explore use of pattern detection tools in

detecting dental claim anomalies  

• 2004 data

• 1,600 Ontario GPs (non-specialists)

• 200,000 claims

• Many examples of anomalous activity discovered

Traditional ToolsTraditional Tools

Rule-based

Strong at identifying claims that match known types of fraudulent activity

Limited to identifying what is known

Typically analyze at the level of a single claim, in isolation

Pattern Detection ToolsPattern Detection Tools

• Analyze millions of claims to reveal patterns– Technology learns what is normal and what is atypical

– not limited by what we already know

• Categorize each claim. Reveal dentists with high percentage of atypical claims– Immediately highlight new questionable behaviors

– Find new large-dollar schemes

– Also identify frequent repetition of small-dollar abuses

Methodology – 3 Methodology – 3 PerspectivesPerspectives1. By Claim e.g. Joe Green’s semi-annual

check-up

2. By Tooth e.g. all work done in 2004 on Joe’s bicuspid by Dr. Brown

3. By General Work  e.g. all general work (exams, fluoride, radiographs, scaling and polishing) done on Joe in 2004 by Dr. Brown

Methodology – 2 Methodology – 2 TechniquesTechniques1. Principal components analysis

2. Clustering

PCA Cluster

Claim by claim Tooth by tooth General work

Methodology – SummaryMethodology – Summary

Principal Components Principal Components AnalysisAnalysis•Visualization technique

•Allows reduction of multi-dimensional data to lower dimensions while maximizing amount of information preserved

•Powerful approach for identifying outliers

PCA: Simple ExamplePCA: Simple ExampleP C A: S am p le D ata

-6

-4

-2

0

2

4

6

-6 -4 -2 0 2 4 6

X -Ax is

Y-A

xis

Lots of variance between points around both axes

PCA: Simple ExamplePCA: Simple ExampleP C A: Ro ta te Axes 45 o

-6

-4

-2

0

2

4

6

-6 -4 -2 0 2 4 6

X -Ax is

Y-A

xis

Create new axes, X’ and Y’: rotate original axes 45º

PCA: Simple ExamplePCA: Simple Example

In the new axes, very little variance around Y’Y’ contains little “information”

P C A: Ro ta ted Axes

-6

-4

-2

0

2

4

6

-8 -6 -4 -2 0 2 4 6 8

X '-Ax is

Y'-

Ax

is

PCA: Simple ExamplePCA: Simple Example

Can set all Y’ values to 0 – ie, ignore Y’ axisResult: reduce to 1 dimension with little loss of info

P C A: Red u ce to 1 D im en s io n

-6

-4

-2

0

2

4

6

-8 -6 -4 -2 0 2 4 6 8

X '-Ax is

Y'-

Ax

is

♦ Original

● Revised

PCA: Identifying Atypical PCA: Identifying Atypical DentistsDentists

PCA: Identifying Atypical PCA: Identifying Atypical DentistsDentists

1. Begin at the individual transaction level

2. Determine the average transaction for each dentist

3. Graph quickly isolates dentists that are outliers

Dentist Average 90th percentile

PCA: Identifying Atypical PCA: Identifying Atypical DentistsDentists

PCA: SummaryPCA: Summary

1. Visualization allows for easy and intuitive identification of dentists that are atypical

2. Manual investigation required to understand why a dentist or claim is atypical

Clustering TechniquesClustering Techniques

•Categorization tools

•Organize claims into several groups of similar claims

•Applicable for profiling dentists by looking at the percentage of claims in each cluster

•We apply two different clustering techniques

– K-Means

– Expectation-Maximization

Clustering ExampleClustering Example

Clustering ExampleClustering Example

Example: Clusters found – by Example: Clusters found – by ClaimClaim 1. Major dental problems

2. Moderate dental problems

3. Age > 28, minor work performed

4. Work includes unbundled procedures

5. Minor work and expensive technologies

6. Age < 28, minor work performed

Clusters: Identifying Atypical Clusters: Identifying Atypical DentistsDentists

1. Begin at the individual claim level

2. Calculate the proportion of claims in each cluster – in total, and for each dentist

3. Isolate dentists with large deviations from average

Clusters: Identifying Atypical Clusters: Identifying Atypical DentistsDentists

Proportion of General Work by ClusterK-Means Clustering

0%

20%

40%

60%

80%

100%

% C

laim

s X

Average 19% 2% 6% 20% 6% 9% 21% 17%

Dent A 0% 0% 0% 0% 0% 49% 0% 51%

1 2 3 4 5 6 7 8

Clustering Techniques: Clustering Techniques: SummarySummary

1. Clustering provides an easy to understand method to profile dentists

2. Unlike PCA, clustering tells why a given dentist is considered atypical

3. Effectively identifies atypical activity at the dentist level

Pilot Project ResultsPilot Project Results

• PCA identified 68 dentists that are atypical

• Clusters identified 182 dentists that are atypical

• 36 dentists are identified as atypical by both PCA and Clusters

• In total, 214 of 1,644 dentists are identified as atypical (13%)

• Billings by atypical dentists were $2.5 MM out of $16.1 MM billed by all dentists (15%)

ExamplesExamplesOfOf

Atypical DentistsAtypical Dentists

Proportion of General Work by ClusterK-Means Clustering

0%

20%

40%

60%

80%

100%%

Cla

ims

X

Average 19% 2% 6% 20% 6% 9% 21% 17%

Dent A 0% 0% 0% 0% 0% 49% 0% 51%

1 2 3 4 5 6 7 8

Analysis: Dentist A is far beyond norms in clusters 6 and 8; both indicate high charges in ‘general dentistry’

What we discovered: Each of Dentist A’s patients is being billed for at least 30 minutes of polishing and 45 minutes of scaling

Dentist ADentist A

Dentist BDentist B

Proportion of Teeth by ClusterK-Means Clustering

0%

20%

40%

60%

80%

100%%

Cla

ims

X

Average 22% 9% 5% 13% 19% 25% 8%

Dent B 5% 20% 21% 0% 0% 0% 54%

1 2 3 4 5 6 7

Analysis : Dentist B is far beyond norms in cluster 7

What we discovered: Frequent extractions and anesthesia Dentist B looks like an oral surgeon, yet is a general practitioner

Dentist CDentist C

Proportion of General Work by ClusterEM Clustering

0%

20%

40%

60%

80%

100%%

Cla

ims

X

Average 23% 77%

Dent C 66% 34%

1 2

Analysis: Very high proportion in Cluster 1, suggesting many high-ticket visits

What we discovered: First, Dentist C performs an inordinate amount of scaling. Second, Dentist C has emergency examinations with atypically high frequency

Dentist MDentist M

Analysis: Dentist M has a disproportionate amount of work beyond the 90th percentile of work by all dentists on all teeth

What we discovered: Dentist M appears to utilize lab work very heavily

More Atypical DentistsMore Atypical Dentists

• Frequent use of panoramic x-rays

• Frequent use of nitrous oxide – including with procedures rarely associated with anesthesia

• Large number of extractions, often using multiple types of sedation

• Very high proportion of claims for crowns and endodontic work

• Individual instruction on oral hygiene provided with very atypical frequency

Dental Fraud SummaryDental Fraud Summary

• Highly effective in identifying dentists with claim portfolios significantly different from the norm

• Enables experts to quickly identify and focus on those dentists with atypical claims activity

Pattern detection:

Disability PricingDisability PricingWithWith

Predictive ModelingPredictive Modeling

• Protects against loss of income due to illness or injury

• Annual incidence rates ≈ 3 per 1,000

• Claim duration ranges from a few days to a few decades

About Group LTD InsuranceAbout Group LTD Insurance

• Base rates: tabular, depend on age, gender, EP and max benefit duration

• Several loading factors

• Demographic info (eg, occupation, area, salary)

• Group characteristics (eg, group size, industry)

• Plan features (eg, benefit %, e’e contributions)

Typical Rate StructureTypical Rate Structure

• Uncover and quantify complex relationships between pricing factors and claim experience

• Maintain wholeness of data

• Improved accuracy vs traditional methods

Why use PM for pricing?Why use PM for pricing?

The ChallengePredictive models can be black boxes, difficult to interpret

The ObjectiveA process to make rate structure explainable to:

Predictive Modeling + Predictive Modeling + PricingPricing

• Management • Sales Force

• Regulators • Customers

The 5 Steps -The 5 Steps -A PM Pricing ProjectA PM Pricing Project

The Five StepsThe Five Steps

1.Set objective

2.Data

3.Select technique and predictors

4.Train the model

5.Validate

Develop a predictive model to:

• Predict claim incidence rates

• Predict claim severity, conditional upon claim being made

for each member of census data

1. Set objective1. Set objective

Required:

• Accurate census data for many groups

• Plan features for these groups (EP, Ben%)

• Characteristics of these groups (region, industry)

• Claim data that often can be linked to census

2a. Pull Data2a. Pull Data

Split Data Into Three Subsets

• Training

• Testing

• Validation

2b. Sample Data2b. Sample Data

• Begin exploring with simpler tools

• Learn key predictors, relationships

• Migrate to more complex tools

• Incorporate exploration learning into model design

• Use test sample to evaluate models

3. Select technique3. Select technique

4. Train the Model4. Train the Model

Pricing factors

Pricing factors

Challenge: Understanding the Model

• Need to extract pricing factors from “black box”

Approach:• Build several models

• Each using same modeling technique

• Each using same data

• Each excluding different predictor

• Compare actual to predicted claim cost across excluded predictor

Understanding the ModelUnderstanding the ModelTraining the Training the

ModelModelTraining the Training the

ModelModel

Looking into the Black BoxLooking into the Black Box

Training the Training the ModelModel

Training the Training the ModelModel

Example: model includes all predictors, consider historic claims by industry

Demonstrates accuracy of model, but provides no information about claim drivers

Industry Actual Predicted

Act/Pred

Finance $1,600 $1,600 100%

Health $1,200 $1,200 100%

Training the Training the ModelModel

Training the Training the ModelModel

Example: model includes all predictors except industry, consider historic claims by industry

Isolates impact of industry from all other factors

Industry Actual Predicted

Act/Pred

Finance $1,600 $1,800 89%

Health $1,200 $1,000 120%

Training the Training the ModelModel

Training the Training the ModelModel

Example: exclude two predictors: region and occ class

Region White Collar

Blue Collar

North East 115% 95%

Mid West 75% 105%

South 105% 115%

Isolates impact by region and occupation class from all other factors

Training the Training the ModelModel

Training the Training the ModelModel

Benefits of Extracting Factors

1. Ability to test importance of predictors, singly or in combinations

• Test predictors that are typically not priced for

• Quantify impact of each predictor or pair of predictors

2. Facilitates use of the traditional Base Rate Approach

Training the Training the ModelModel

Training the Training the ModelModel

Base Rate:

• Function of age, gender, EP, max duration

Loadings:

• For other demographic info, group characteristics, plan features

• Determine by looking inside black box

• Company can adjust final factors

Base Rate ApproachBase Rate ApproachTraining the Training the

ModelModelTraining the Training the

ModelModel

Base Rate:

Base rate = 1.20

Loadings:

Base Rate ExampleBase Rate Example

Ben% Region Industry

Salary Contrib Fact X

60% NE Health $60K Y 10.2

+5% -5% +15% -5% +10% -3%

Age Gender EP Max Dur

45 Male 180 days Age 65

Training the Training the ModelModel

Training the Training the ModelModel

Base Rate = 1.20

Loadings = {+5%, -5%, +15%, -5%, +10%, -3%}

Rate = 1.20 *1.05 *0.95 *1.15 *0.95 *1.10 *0.97

Final Rate = 1.395

Base Rate ExampleBase Rate ExampleTraining the Training the

ModelModelTraining the Training the

ModelModel

• Use excluded data for validation of model

• Compare actual vs predicted claim costs

• Ideally, compare predictive modeling approach to traditional approach

5. Pricing Validation5. Pricing Validation

• Improved accuracy vs traditional methods

• Capture effects of interaction between several predictors

• Able to isolate impact of any predictor or pair of predictors

• Able to test new predictors for impact on claim cost

• Can maintain existing Base Rate structure

Advantages of PM: Pricing

Questions?Questions?