Predictive Modeling for Workers Compensation Claims AASCIF Super Conference Kirsten C. Hernan Deloitte Consulting LLP October 4, 2012
NOTICE: THIS DOCUMENT IS PROPRIETARY AND CONFIDENTIAL This document is protected under the copyright laws of the United States and other countries. This document contains information that is proprietary and confidential to Deloitte Consulting LLP and Deloitte Development LLC and shall not be disclosed to outside parties, duplicated, or used in whole or in part for any purpose other than to evaluate Deloitte Consulting LLP. Any use or disclosure in whole or in part of this information without the express written permission of Deloitte Consulting LLP is prohibited. -1-
Agenda Analytics and Predictive Modeling Overview Severity Modeling for Claims Business Applications Questions & Answers -2-
Advancing the Claims Process Effective claims organizations leverage information and technology to gain insight into their operations and performance. to achieve superior service, optimal loss outcomes, and efficient operating costs. Breaking through the barriers Legacy systems and infrastructure challenges Perceived poor data quality and completeness Action Numerous reports produced, few bring value Limited internal analytic capabilities Data Extracting detailed data stored in internal and external systems and databases Information Transforming data into information that can be analyzed and understood -3- Insight Exploring and interpreting trends and performance to understand drivers Converting insights into business actions that drive results Continuous Improvement
The Evolution of Analytics Level of capabilities 1 2 3 Loss and expense savings Actuarial Analysis Data Analysis Predictive Modeling # Analytics Approach Description Translation 1 Actuarial Analysis Periodic analysis to quantify overall losses costs and trends based on data triangles and in some cases limited claim detail Projecting ultimate losses and highlevel drivers of trend 2 3 Data Analysis Predictive Modeling Data drill down to identify both traditional and non-traditional cost drivers with focus on readily available data Applies data mining and statistical algorithms to predict outcomes leveraging a variety of internal and external data sources -4- Finding meaningful patterns in your data Predictive models that generate a score and drivers of exposure
Claims Predictive Model Design Predictive modeling is the application of data mining and statistical techniques to effectively predict and segment claim outcomes by level of exposure, by transforming internal and external claim characteristics into model outputs to enable business actions that drive results. Model Inputs Several hundred internal and external characteristics are tested to identify the 50-100 with greatest predictive power Claim Segmentation Curve Data Mining & Statistical techniques Sample Model Equation w 1 (Claimant Age) + w 2 (Dist_H_W) +w 3 (Emerg_ Rm) + w 4 (Job Function) + w 5 (CoMorbidity) + w 6 (Report_lag) +. Outcomes Model Outputs John Smith 1 Circle Ave. Anytown, NY 92 Reason Messages: Multiple co-morbidities Claim history Emergency Room Distance from work -5- Low Claim Complexity High
Agenda Analytics and Predictive Modeling Overview Severity Modeling for Claims Business Applications Questions & Answers -6-
Loss Segmentation Typically, 20% of claims represent 70-80% of loss costs Quickly and prospectively identifying claim exposure can enable proactive management strategies that can help drive better business results. 100% 20% 75% Claim Assignment Fraud Detection 50% 80% Claim Escalation 80% Competitive Advantage Reduced Loss Costs Medical Management 25% Business Metrics Claim Escalation 0% Claim Counts 20% Loss Costs -7-
Data is the Key By combining internal data with public external data, enhanced segmentation can be achieved. Claim Data Claimant Data Type of injury Injury timing Report lag Insured Data Tenure Type of work and job level Marital status Billing Policy and coverage Data Experience data Policy Data Synthetic Data Possible data sources Medical Data Performance metrics Number of employees Vendors/contractors usage Location specifics Complaints External Data Census /demographic Household Insured financial strength Industry loss data Legal environment -8- Treatment patterns Prior claim experience Pre-existing conditions Prescription usage
Providing Insights Beyond Injury Group Injury Group segmentation* provides further insight than traditional methods enabling segmentation of claims within like injuries and allowing other severity characteristics to come through. $160,000 $140,000 $120,000 $100,000 $80,000 $60,000 $40,000 Ave Severity: Sprain Strain Shoulder Ave Severity: Sprain Strain Neck & Back Ave Severity Contusion Excl Back & Shoulder $20,000 $0 1-10 91-100 Contusion Excluding Back & Shoulder Sprain Strain Shoulder Other * Patent Pending -9- Sprain Strain Neck & Back
Capturing the Dynamic Nature of Claims Advanced claims models recognize the dynamic nature of claims and provides through-time updated scores driven by new information (e.g., medical bills, pharmacy data). Aggregate Number of Office Visits Relative Projected Loss Cost 100% 80% 60% 40% 20% 0% -20% -40% -60% 0 1 2 3 4 5 6-10 11-20 21-30 31-60 Over 60 Month 3 Month 6 Month 12 Not all data is available at each stage in the claim lifecycle Characteristics change through time Expectations around certain variables can vary across time and injury types. Advanced modeling capabilities are required to address complexity arising from time-dependent predictive information. -10-
Enhanced Segmentation can be Achieved through Predictive Modeling Claims predictive models have the ability to consistently differentiate high cost claims from low cost claims from first report and through the life of the claim. 150% 100% 50% 0% At first notice claims scoring above 90 were over 3 times as costly as those scoring below 10 With 1 month that segmentation was enhanced to over 10 times -50% -100% Best 10% Worst 10% FNOL Mth 1 Mth 3 Mth 6-11-
Agenda Analytics and Predictive Modeling Overview Severity Modeling for Claims Business Applications Questions & Answers -12-
Claims Management With Claims Predictive Modeling An integrated Severity Predictive Modeling solution transforms the allocation of claim resources, enabling more efficient claim management, improved claim severity estimation, enhanced fraud detection and shorter claim cycles. Re-Scoring Loss 1 Event Assign Claim Investigate/ Evaluate Establish Reserves Escalation? Manage Claim Issue Payments Close First Report of Injury Claim Model Triage Low Touch Fast Track 2 SIU Investigation Subrogation / Salvage Litigation Management 3 Medical Management 4 Oversight 1 Triage & Assign Claim 2 SIU Investigation Key Applications 3 Medical Management 4 Oversight -13-
Transforming Data into Action Data and model insights are translated into business actions through the incorporation of model output into business rules. Score Business rules Claims data (e.g. claim, medical, claimant) 92 Fracture Foot Reason messages Assign all claims with score >80 to Tier 4 adjusters Internal data (e.g. Insured ) CPM Distance from home to work indicates higher exposure Insured has negative financial management experience Number of prior claims indicates a higher exposure Refer all claims with score > 75 and 3+ red flag reason messages to SIU Investigator Impact variables External data (e.g. census, demographic, insured, financial strength, household) 45 miles to work 5 prior claims -14- Poor insured financial stress score Accident on weekend Assign nurse consultants to all claims scoring > 90 in Injury Group Fracture Foot
Output can assist with ongoing investigation, use of specialty resources First Notice of Loss Summary The claimant, who was a four year employee, worked as a heavy machine operator. The claimant (44 years old) suffered a back strain after a slip and fall. Return to work date unknown. Employer did not question legitimacy of claim. Translating Model Outputs High claim severity score indicates need for experienced claim resource Prior loss history prompted review of past claims with handling adjusters prior to initial claimant contact Model score, claim history, accident characteristics, and distance variable triggered an automated SIU referral Business Actions Assignment to senior claim adjuster Adjuster took the claimant s recorded statement to document the accident and injury details SIU Investigator screened and accepted case referral When initiating contact at the claimant s residence, the investigator was referred to a nearby bar operated by a friend of the claimant Claimant was observed working and, once questioned, indicated he would withdraw the claim Model Score: 87 (High Severity) Results SIU field investigation revealed claimant was not disabled The claim was denied with no payouts made No additional follow up by the claimant or attorney -15-
Claims Organizations are Realizing Significant Benefits Based on the results of the post go-live audits and ongoing measurements, the client reported achieving significant benefits from implementing the End-to-End Workers Compensation Claims Predictive Model Solution, beyond initial targets and expectations. Claim Routing & Assignment Right claim, right resource Improved routing to auto-adjudication Increased triage consistency through automation 25% redeployment of supervisory resources Fraud Detection Reduced lag time of SIU referrals Improved mix of claims referred to SIU Deterrence of soft-fraud 10% improvement in SIU caseload Over 8% Reduction in Loss Cost Medical Management Capability Growth Prompt assignment of nurses on those cases that need it most Enhanced the collective knowledge, self sufficiency and skills of the project team Integrated behavior issues into nurse members and claims staff assignment Cost effective use of field case management 7% improvement in nurse managed claims -16- Demonstrated ability to close claims faster and at a reduced overall loss cost leads to competitive market advantage Improved client satisfaction and strengthened the relationship/brand
Questions & Answers -17-
Deloitte Consulting LLP 1700 Market Street Philadelphia, PA 19103 USA Kirsten C. Hernan Tel: (215) 246-2391 khernan@deloitte.com Member of Deloitte Touche Tohmatsu -18-