Decision Analytics and Modeling

The gap in real-world evidence has emerged as a critical challenge to drug development and market access strategy. Complex decisions have to be taken from early stages to launch, reimbursement and pricing, that should take into account the real world value of the medicines and health products. A series of scenarios must be explored in each situation, and defended to decision makers and payers.

Through a suite of dedicated and proven methodologies, our Real-World Analytics team can leverage and integrate all the relevant information available at any stage of development and market access to estimate or predict effectiveness in a given context and to inform strategic decisions for drug evaluation and evidence generation.

LASER uses the best of modeling, simulation, Bayesian statistics and epidemiological sciences to build effectiveness, relative effectiveness and real-life benefit/risk arguments specific to each country. All predictive models are used.

  • Bridging-to-Real World Studies™: this proprietary study concept combines advanced predictive modeling with targeted real-world data-collection.
  • Strategic Development Analytics: early decision modeling & optimal clinical program design, surrogate endpoint validation & epidemiological forecasting
  • Economic Optimization: therapy sequence modeling & value-based pricing optimization
  • Access Data Platforms: clinical and observational data are integrated into a platform to optimize evidence generation and real world impact assessment
  • Quantitative Evidence Synthesis: These powerful methods provide very useful information to support HTA submissions, value dossiers, as well as internal decision-making for drug development.

The combination of our cutting-edge Modeling & Simulation Methods as well as advanced Epidemiological Expertise allows us to provide our customers with a unique service.

Leadership

Billy Amzal, UK & France
Hélène Karcher, Germany

Directors & Consultants

Sumeet Bakshi, UK

Keep in touch with us

Bridging to Real World

Real-world clinical studies take time and resources to do; they cannot be conducted in every country for every drug at all stages of drug lifecycles. Yet clinical development data and many, however partial, real-world data sources may still be available to inform on the expected effectiveness or relative effectiveness.

An unrivalled option to anticipate drug real-world value

Bridging studies combine advanced predictive and integrative modeling with targeted clinical and real-life data analyses. Prospective or retrospective data collection can be designed in an optimal way leveraging all evidence available at present. Such studies can be used to bridge from efficacy to effectiveness, from country to country, and from one population to another. Bridging models allow individual patient simulations over time that mimic the dynamics of health outcomes, prescriptions and adherence over time. They rely on a structured analysis and quantification of the factors interacting with efficacy in real-life and their interaction with each other and over time. These effectiveness drivers include adherence, prescription patterns, treatment pathways, and patient characteristics. Data from electronic healthcare databases, our PGRx information system, disease registries or real-world studies in other countries are summarised into Access Data Platforms.

The design and conduct of a bridging study requires advanced pharmacoepidemiological and statistical capabilities such as Bayesian hierarchical modeling, copula models, stochastic time-dependent processes and continuous Markov chains.

Bridging  effectiveness models can be complemented by a cost layer to construct a real-world cost-effectiveness model providing more accurate outputs than efficacy-based models. These models can also be used to support risk-sharing agreements and value-based pricing. 

Strategic development analytics

LASER offers superior analytics to evaluate and document the anticipated real-life benefit, risk and cost of new interventions and to support internal decisions (market access strategy for products, port-folio prioritization, trial design), evidence building for a particular product and applications to authorities for approval and reimbursement.

We use a variety of model types as appropriate:Markov/cohort,decision analytic,Bayesian,patient-level models, discrete-event/Monte Carlo simulations, and typically include deterministic and probabilistic sensitivity analyses. Our Modeling & Simulation methods are state of the art.

Our services include:

Early decision analytic models

LASER designs customized disease models to explore scenarios and recast a future product’s value into evolving therapeutic landscapes. Our models help position products but also identify uncertainties and gaps in evidence when there is still time to fill them. Models are built on structured clinical and observational data, clinician consultations, appropriate early mixed treatment comparisons, etc., and typically include comparisons of clinical outcomes, market share or cost-effectiveness/consequences/utility for different scenarios.

Strategic optimal clinical program design

We provide forecasting support to evaluate the probability of success for our clients’ individual trials or clinical program under different scenarios. Our simulations combine:

  • Clinical insight, including forecasting of clinical endpoints and real-world outcomes from biomarkers and surrogates
  • Sound and transparent statistical design, including variability and uncertainty estimation

We strive to help decisionmakers gain confidence in their choice of target population, sample size, choice of primary and secondary endpoints for optimal trial results.

Epidemiology forecasting models

These models are a subtype of early decision analytic models that consider the macro-level evolution of a disease and associated consequences to best anticipate changes due to evolution in therapy standards and populations for a particular disease and setting.

Surrogates of clinically relevant endpoints are increasingly often relied upon to assess benefit-risk and drive decisions on approval and reimbursement of new therapies.

Surrogate endpoint validation

We propose state-of-the-art surrogate validation to justify their use instead of clinical endpoints (in accordance with the most recent IQWiG or NICE guidelines)

Economic optimization models

Therapy sequence models

LASER offers the possibility of assessing sequences of treatments in silico. Our models are tailored to the indication of interest. They can be based on existing clinical data for sequences – such as time to clinical endpoints or frequency of adverse events – or on expert-driven assumptions on these parameters. The estimation of outcomes for each treatment sequence can include:

  • Total clinical benefit, such as progression-free survival and overall survival,
  • Total cost, for example including drug cost and cost of treating adverse events,
  • Patient distributions across therapies at each treatment line, and
  • Required volume of each drug under several sequencing scenarios

Our models are transparent and make it easy to assess increasingly complex therapeutic environments with multiple possible treatment sequences.

Value-based pricing optimization

We provide a transparent way of assessing and weighing risks of non-reimbursement while maximizing earnings to back up negotiations towards garnering or maintaining therapy reimbursement.

Our price models are built on relevant historical data and guide the choice of the optimal price/population/funding conditions for the given therapy, in reducing uncertainty and narrowing down the range of options to choose from.
Our simple graphical representation of pricing implications makes it easy to visualize options.
See also Economic Modelling to support

Modeling & Simulation Methods

We use a variety of model types as appropriate: Markov/cohort, decision analytic, Bayesian, patient-level models, discrete-event/Monte Carlo simulations, and typically include deterministic and probabilistic sensitivity analyses. Our strengths are:

Thoughtful definition of the modeling framework and scope

  • We support our customers in assessing all aspects of their business needs (technical, strategic, scientific and short and long-term) and tailor the model to meet them.
  • We advise – in an unbiased way – on possible data sources and on the use of indirect, derived data to feed models

Transparent model building process

  • The client always knows exactly what is being done and has full control over model deliverables. We explain and share our codes upon request.
  • We are expert in most programming languages, including Excel/VBA, R, SAS, STATA, SPSS, TreeAge, Winbugs, C/C++, Java and Matlab, and therefore tailor programming to scope and to our clients’ personal preferences.

Advanced and broad modeling skills

  • The most appropriate techniques are suggested, which can be very simple or more elaborated. As a company with advanced level in all modeling and statistical techniques, we never try to sell our “one favorite” technique or model type

Integrated solutions

  • We can combine our modeling work with any other activity (study, submission dossier, MCDA, strategy consulting, polished web-based interface, training) to create the best possible solution to our customer’s need.

Access Data Platforms

Clinical and observational data can be integrated into a data platform to be used and developed to optimize evidence generation planning and also for bridging studies. This so-called Access Data Platform is a convenient toolbox for multiple use. LASER has access to and extensive experience with a large series of electronic medical records (eMR) and electronic health care databases (eHCD) in the US, Europe and Canada and also provides a unique resource for disease studies through it’s PGRx Information System.
The analysis of these databases benefits from the advanced pharmacoepidemiological and statistical expertise of our real-world research teams.

Quantitative Evidence Synthesis

To ensure our work is of the highest quality, we employ all available and proven methods, eg:

  • Meta-analysis
  • Publication bias assessment
  • Heterogeneity assessment
  • Between groups homogeneity testing (categorical and continuous data)
  • Multivariate analyses
  • Estimation of overpopulation dispersion
  • Survival analysis including cross-over effect analysis
  • Generalized linear models (linear regression, logistic regression etc.) including prediction analysis
  • Network meta-analyses
  • Bayesian hierarchical models
  • Individual and trial level surrogate validation

These powerful methods provide very useful information to support HTA submissions, value dossiers, as well as internal decision-making for drug development.