leveraging anonymous patient level data to identify...
TRANSCRIPT
DATA INPUTS
Leveraging Anonymous Patient Level Data to Identify Undiagnosed Rare Disease Patients
Jennifer RawdingSr. Consultant, Commercial Effectiveness (302) 463-8411; [email protected]
Julieanna GubitosaSr. Manager, Commercial Effectiveness (610) 996-6760; [email protected]
Jonathan WoodringPrincipal, Commercial Effectiveness (201) 421-6472; [email protected]
BACKGROUND
METHODOLOGY
CONCLUSION
SHS combines anonymous patient level data with disease state knowledge and machine learning techniques to identify potential rare disease patients prior to diagnosis. This enables identification of patients and physicians in “real time,” such that interventions are
timely, targeted and effective.
Extremely small populations, complex symptomatology, and lack of physician awareness can make identifying and diagnosing rare disease patientsextremely difficult.
Challenge: Identifying the appropriate time to engage a physician is paramount for diagnosing rare disease patients
SHS Solution: Real-time Promotional Analytics Identifying appropriate medical education through daily or weekly reports that monitor changes in patient level claims data patterns.
TriggerIdentification
Trigger Design
TriggerImplementation1 2
Deduction Based
Induction Based
Historical Lookback
Tracking & Impact
Deduction Based
Collaborate with brand, marketing and clinical teams to identify events that could potentially trigger greater receptivity to promotion
Induction Based
Data mining to identify events that could potentially trigger greater receptivity to promotion
Historical review of the data associated with the trigger to…
• Estimate the opportunity related to the trigger
• Identify the ideal outreach timing based on the trigger (given the trigger event and physician action)
• Determine the optimal cadence/frequency of the trigger
• Incorporate triggers into sales planning and deployment processes
• Post-implementation impact assessment at a trigger level to estimate the lift in script volume due to the trigger
• Tracking and reporting metrics associated with the triggers on a regular basis.
Build & Deploy
Operational Assessment
Rare Disease Patient Journey
Deduction Based Trigger Identification
Possible Additional Layers:• Age • Demographics
Seizures + Diagnostic testing
Juvenile Hyper-cholesterolemia+
Hepatosplenomegaly
Pre-Diagnosis Treatment
Primary Symptoms
Secondary Symptoms
Comorbidities
Misdiagnoses
Diagnostic Testing
Genetic Screening
Rare Disease Diagnosis
Diagnosis
Break in Therapy
2 Primary Symptoms + 2 Secondary
Symptoms
Beta Glucosidase Assay + Organ
Swelling
New Rare Disease Diagnosis
Pancreatitis + Metabolic Panels
Xanthoma + Cholesterol Test
Acromegaloid + Hepatosplenomegaly
+ Acanthosis
Bowel Surgery + Chronic Diarrhea +
Parenteral Nutrition
Payer Rejection
Competitive Treatment Initiation
1 Trigger Design2
Preliminary Trigger List
Analyze historical data associated with each trigger to optimize trigger definition to balance…
Trigger Volume Trigger Timing Trigger Delivery Cadence
Optimizing trigger volume by adjusting specificity/thresholds
Optimizing timing of the trigger based on trigger event and physician action
Optimizing delivery cadence of the trigger based on territory level volume & capacity
Physician Action
Trigger Event
Operational Assessment
Scanning SHS patient claims data on a continuous basis for the defined triggers…
…and delivering daily, weekly, or monthly trigger files
Relay triggers to personal and non-personal channels (Medscape, Doximity, Epocrates) and non-selling teams (MSLs, Nurse Educators, Case
Managers) on a continuous basis at a defined frequency…
Sales Force
Nurse EducatorsCase Managers
Non-personal Channels
Medical Science Liaisons
Build & Deploy
Trigger Implementation3
Anonymized patient level data can and is helping increase the awareness,
diagnosis, and treatment of patients in the rare disease space.
The scale of the SHS foundational data assets, and the strength of the deductive and inductive
algorithms deployed, are assisting leaders in the rare disease space to increase important medical
education, appropriate diagnosis, and life-changing treatment for those suffering from rare disease.
RESULTS
2%5%5%5%5%5%
7%9%9%
18%
0% 10% 20%
PED GASTROENTEROLOGYFAMILY MEDICINE
HEMATOLOGYPATHOLOGY
INTERNAL MEDICINEPEDIATRICS
PED HEM/ONCMED ONC
GASTROENTEROLOGYHEM/ONC
Physician Volume by Specialty
0102030405060708090
100
Ohi
oSo
uthe
rn C
AN
J/PA
New
Yor
k Ci
tySo
uth
Flor
ida
Loui
svill
eSo
uth
Texa
sM
I/In
dian
apol
isPa
cifi
c N
orth
wes
tAr
izon
a/U
tah
Nor
th T
exas
Nor
ther
n N
ew…
Min
neso
taSo
uth
Cent
ral
Atla
nta/
Colu
mbi
aD
CSo
uthe
rn N
ew…
Ups
tate
NY
Nor
ther
n CA
Gre
ensb
oro/
Ch…
Colo
rado
KS/M
OIo
wa
Haw
aii -
Hyb
rid
Triggers by Territory• Avg. Triggers / Territory: 27• % Triggers Aligned: 15%
Avg.
67%
34% 43% 46%
19%40%
29%39%
14%29%
19%42%
4%
1%1%
1%
3%
21%
20%
25%
3%
26%
18%
33%
6%
29%27%
21%
44%
13%18%
10%
32%
21%
29%
3%18%
33% 27% 22% 30%17% 23%
13%
49%
16% 21% 12%
Genetic Testing + Organ Swelling
Xanthoma + Cholesterol Test
Symptoms + Organ Swelling
Misdiagnosis
2 Major Symptoms + 2 Minor Symptoms
Pancreatitis + Lipid Panel
New Diagnosis
Competitive Treatment
Any Treatment + Adverse Events
Break in Therapy
Payer Rejection
Treatment + Biomarker
1
Induction Based Trigger Identification
1
2
3
Leverage Machine Learning Techniques to Identify Existing Rare Disease Patients• Collaborate with brand and medical teams to identify ‘ideal patients’• Derive all relevant variables across Rx, Tx, Px, Dx dimensions and
demographics across all SHS core data assets• Examine univariate distributions for training and hold out sample
Conduct Patient Look-Alike Modeling• Leverage advanced analytics and bootstrapping techniques • Score entire SHS (eligible) patient database • Validate model and assess model diagnostics
Patient Aggregation and Physician Attribution• Aggregate scored patient database to the physician level• Decide upon single physician attribution or allow multiple attribution• Initiate physician level targeting
Identify existing rare disease patients and score the entire SHS patient database for these characteristics, aggregate to the physician level based on the most recent data available…
1
TRIGGER when patient presents with unique
combination of symptoms
TRIGGER when physician orders tests for suspected
condition
TRIGGER when the patient is diagnosed with the
condition
Making diagnosis and treatment decisions for patients
TRIGGER when payer information for a patient
changes
HAE MarketIdiopathic Thrombocytopenic Purpura~15% increase in new patient starts on the client’s drug in the test district versus an equivalent control district
Brand’s market share among the triggered patient population was measured to be double as compared non-triggered patient base
LSD Market
Off the triggers that are delivered to Reps, ~34% got initiated on therapy; On the non-triggered patient set, only ~16% got initiated on therapy
Trigger ImplementationTracking &
Impact
Patient-centric longitudinal database enables an in-depth understanding of the patient journey
IDV® captures & connects patient, physician, pharmacy and hospital data from healthcare payment processing
Aver
age
year
s of
patie
nt
exist
ence
in
IDV
IDV highlights various Patient Touchpoints