A reusable predictor approach

Cassis develops predictive decision-support tools for complex pharmaceutical and healthcare decisions.

The work addresses a practical question:

Can we identify changes in clinical, commercial or market-access risk early enough to improve the decision?

Pharmaceutical organisations generate large volumes of clinical, safety, economic and market data. The difficulty is not usually a lack of information, but converting fragmented evidence into a forward-looking decision.

Cassis develops predictors for defined decisions across the product lifecycle, including:

• health technology assessment;
• pricing and reimbursement;
• market access;
• pharmacovigilance;
• medication-related harm;
• clinical reasoning;
• patient risk and treatment response.

The problem

Important decisions are often based on:

• evidence distributed across several sources;
• retrospective analysis;
• population averages;
• inconsistent interpretation;
• uncertain assumptions;
• limited modelling of alternative decisions.

This can delay recognition of clinical risk, weaken pricing or reimbursement strategy, and reduce the ability to anticipate how regulators, HTA bodies, payers or clinicians may respond. The objective is to estimate how a defined outcome is likely to change before the decision is made.

Why Cassis is different

Cassis begins with the decision, not the AI technology.

Each predictor defines:

• the decision to be improved;
• the outcome to be predicted;
• the available evidence;
• the relevant uncertainty;
• the possible actions;
• the consequences of error.

This supports a shift from:

• retrospective analysis to forward prediction;
• static evidence to changing trajectories;
• population averages to patient or market-specific estimates;
• generic dashboards to decision-focused outputs;
• single-point estimates to quantified uncertainty.

The aim is to improve judgement under uncertainty.

Pharma applications

Potential applications include:

• predicting HTA outcomes;
• modelling pricing and reimbursement scenarios;
• predicting payer or formulary response;
• modelling the effect of different medicine value propositions;
• forecasting treatment uptake;
• identifying emerging safety signals;
• estimating medication-attributable vulnerability;
• predicting patient deterioration or treatment-related harm.

Reusable architecture

Although each predictor addresses a different problem, the development method is consistent. Cassis has developed:

• defined use cases;
• predictor logic and mathematical specifications;
• interactive HTML simulators;
• synthetic cohorts and scenarios;
• patient-specific and population-based models;
• risk visualisation;
• decision outputs for clinical and commercial users;
• documentation for technical, clinical and commercial review.

The economic value lies in the reusable method. Once the core architecture, data standards, explainability and decision interface are established, additional applications can be developed more efficiently.

Commercial proposition

Cassis is seeking pharmaceutical and life-science partners interested in evaluating, licensing or co-developing predictor applications. The value lies in:

• a repeatable development method;
• working simulations;
• interpretable models;
• applications across clinical, safety, pricing and market-access decisions;
• the ability to test scenarios before committing resources;
• the potential to extend one validated application into a broader predictor platform.

For more information

Cassis welcomes enquiries from pharmaceutical companies, investors and strategic partners interested in predictive decision support across the product lifecycle.

To request a private demonstration or discuss a feasibility study, contact mike_tremblay@skythunder.net.