Bayesian Pipelines
Pipeline physics created a new model of phramaceutical pipelines. This model describes each phase and its downstream impact by using Bayes' law. For example, Bayes' law describes:
- How the management of preclinical tests affect phase I trials.
- How the management of phase I trials affects phase II trials.
- How the management of phase II trials affects phase III trials.
The Bayesian pipeline model makes different predictions than other pipeline models, including those of well-known and widely used pipeline analyses. Pipeline Physics believes our Bayesian model is correct and previous models are misleading executives and harming their pipelines. Executives should use our new Bayesian pipeline model.
Pipeline Physics is pairing its model with a new statistical analysis to produce incisive metrics. The following definitions help to introduce the metrics:
- Marketable compound: a compound that is safe and effective.
- Unmarketable compound: a compound that is unsafe or ineffective.
Using these definitions, Pipeline Physics' statistical analysis estimates the following qualities of each discovery and development phase:
- Base rate: The percentage of compounds evaluated by a phase that are marketable.
- False-positive rate: The percentage of unmarketable compounds a phase mistakenly advances downstream.
- False-negative rate: The percentage of marketable compounds a phase mistakenly cancels.
- Resolution: A phase's ability to distinguish marketable from unmarketable compounds.
- Impact of throughput: How a phase's throughput (success rate) affects its false-positive and false-negative rates.
The need for these statistics is well-known, but the statistics have never previously been produced because of a pesky problem. One never learns whether a canceled compound is marketable or unmarketable, and without this classification one could not estimate the above metrics. (See Table 1.)
| NDA result | |||
| Success (marketable) |
Failure (unmarketable) |
||
| Development compounds |
Advance to NDA | true-positive | false-positive |
| Cancel during development |
true-negative or false-negative |
||
Having solved the problem of unclassified compounds, Pipeline Physics is performing the following analyses:
- Estimating each phase's base rate, false-positive rate, false-negative rate, resolution and how a phase's throughput (success rate) affects its false-positive and false-negative rates.
- A new sensitivity analysis to indentify the best opportunities for raising pipeline productivity.
- Determining if phase II's actual false-positive and false-negative rates are higher than phase II's trials were desiged to produce? If so, why?
- Estimating the maximum resolution previous clinical trials could produce.
- Estimating the resolution that pharmaceutical companies evaluation and selection methods actually produced.
- Revealing how popular metrics and practices - such as NPV, scoring models and portfolio optimization - affect each phase's resolution.
- Benchmarking of pipeline management best practices.
- Ranking pharmaceutical companies by their clinical trials' resultion and by their pipeline management.
- Ranking CROs by their clinical trials' resolution.
If we can secure sufficient data, Pipeline Physics will perform these analyses for categories of compounds, including therapeutic areas, new molecular entities (NME) and non-NMEs, and large and small molecules.
To learn more about our Bayesian pipeline model and our pipeline analyses, kindly contact Gary Summers at Pipeline Physics.
© 2015 Pipeline Physics. All rights reserved.