Advanced Analytics

Access critical patient insights

Advanced Analytics focuses on deploying innovative methods grounded in the foundations of statistics and econometrics to generate fit-for-purpose health economic and real-world evidence. Our team develops a resilient data and analytics strategy imbued with data-driven critical thinking to support your goals. Taking a data-agnostic approach, Parexel combines unique data access, linking capability, and advanced analytics toward that end. We work in partnership with a large network of vendors, linking to the data sources most relevant for the project, and creating a common pool of data to be mined for critical patient insights using these innovative methods.

Indirect treatment comparison / Meta-analysis

Evidence-based healthcare decision-making requires the assessment of competing interventions. However, head-to-head comparisons are in most cases not available or feasible in randomized clinical trials. Indirect treatment comparisons (ITCs) and network meta-analyses of the therapies of interest allow for the estimation of the relative safety and efficacy of these interventions. Combining the results of multiple trials, a network of evidence may be empirically assessed to identify the best treatment(s). Advanced methods such as maximum likelihood meta-regression, simulated treatment comparisons, and match-adjusted indirect comparisons provide tailored options to address specific data challenges including single-arm trials and incomplete networks.

Advanced Parametric Methods

Standard parametric survival models are often limited in modelling hazard functions that follow more complex patterns. In cases where there is evidence that the hazard function of a therapy has important changes over time that cannot be reflected by standard parametric distributions, it is necessary to explore other approaches. Our team is well-versed in applying advanced parametric methods such as Mixture Cure Fraction models where there is evidence that a proportion of the population treated with the intervention can be considered to be ‘cured’ (the ‘cure fraction’), with similar mortality as that of the general (i.e., background mortality). Alternatively, Bayesian Multi-parameter Evidence Synthesis modelling synthesizes RWD with RCT data to generate realistic long-term extrapolations and by incorporating general population/registry survival rates.  These techniques allow our clients to successfully engage earlier with HTAs using relatively immature data with shorter patient follow-up.

Predictive Analytics, machine learning, and artificial intelligence

Life sciences companies today can leverage an extraordinary array of technologies to support their work in evaluating the potential of their products and bringing them to market. Advancements including natural language processing (NLP), machine learning (ML), artificial intelligence (AI), with predictive analytics provide critical patient insights from big data, whether arising from clinical trials or gathered from literature or real-world sources. For instance, effectiveness research has focused on average treatment effects, but the benefit of the treatments often differs across patients. Further, there is a growing trend towards targeted therapies/precision based on molecular and immunologic status, improving overall survival in patients with advanced disease by providing optimal treatment. Our team of experts use predictive analytics to drive patient identification and support protocol optimization and value-based discussions with answers to questions like:

  • Which set of prognostic factors/patient characteristics will accurately predict outcomes?
  • Can we use these predictors to estimate individual patient-level probabilities for single outcomes/multiple outcomes simultaneously?
  • Can we identify the variables that are most informative to collect for future patients?
  • How do we define optimal patient response profiles to flag those who may be at increased risk for adverse events or improved outcomes?
  • How can we combine the above information to facilitate clinical decision-making?

We focus on demystifying machine learning and providing transparent, robust, solutions with approaches such as Bayesian Networks machine learning, targeted machine learning, and super learners.

Data Visualization and bespoke programming support

A key trend in the evolution of the healthcare ecosystem is the development of digital solutions with data visualization tools/apps and platforms to facilitate multi-stakeholder decision-making.  To this end, our team routinely develops indication-agnostic tools including R-Shiny early economic modeling platforms, NMA tools, cost-effectiveness platforms, budget-impact tools, and risk-sharing tools. Our experts also provide bespoke programming support in dedicated roles to staff internal client solutions.

 

We are always available for a conversation.

*  

We are always available for a conversation.

Submitting...

Communication Preference

Communication Preference