Data and Model Disclaimer

Panoptic Bio, Inc. publishes numbers about clinical trials and about the companies running them. Some are copied from public registries. Our own models produce others. Analysts here type some in by hand. Other people publish some of them, and we show those as they came. And some are placeholders on screens that are still design previews. This page tells you which is which, and what you should not do with any of them.

Nothing we publish is investment advice, and nothing we publish is medical advice. Do not make a decision that matters on our numbers alone.

This page covers panoptic.bio, trialterminal.bio, tower.panoptic.bio and finance.panoptic.bio.

The Finance tool is a preview, and its figures are placeholders

The Finance tool at finance.panoptic.bio, and the same pages at /finance on our other sites, is a design preview. Its financial figures are not real.

All of the following are placeholder values typed into the page. None come from a market data feed, because the product is not connected to one:

• share prices, and 52 week highs and lows
• shares outstanding, float, short interest and average volume
• cash, quarterly burn and enterprise value
• institutional ownership
• analyst price targets and analyst ratings
• company market values, per drug valuations, peak sales figures and pipeline scores

The catalyst calendar on those pages is a separate case. Its 38 FDA dates are written into the page rather than drawn from our calendar feed, so they do not update and we cannot tell you they are current. That is a different problem from the figures above: the dates may well be right, but nothing keeps them right.

These figures sit beside the names of real listed companies, including Vertex, Arrowhead, Beam, BioMarin and Editas. They are not those companies' figures, they are not current, and no model produced them. Do not read them as a view on any of those companies, and do not trade on them.

The Finance pages carry no warning of their own. Do not take the absence of a warning as a sign that a figure on them is real.

Where the trial data comes from

Trial records come from public clinical trial registries. In the production Trial Terminal that means ClinicalTrials.gov and the EU Clinical Trials Information System. Our staging build also holds records from the World Health Organization family of registries. Where a record shows a registry name beside it, that name is the registry the record came from.

We copy what those registries publish. We do not correct it and we cannot confirm it. A trial record is what its sponsor filed, so it can be incomplete, out of date, or wrong.

Drug details, such as chemical structures and drug classes, come from ChEMBL, PubChem, RxNorm and openFDA. Disease terms come from MeSH, courtesy of the U.S. National Library of Medicine, and from MONDO and Orphanet, both used under CC BY 4.0.

Probability of success

The probability of success shown against a trial is a model estimate. We trained a machine learning model that reads a trial's public registry record, including the phase, the condition, the drug and the eligibility text, and estimates the chance that the trial moves past its current phase.

Read it with these limits:

• It is a statistical estimate, not a forecast of what will happen to this particular trial.
• It is not a statement about any patient, any treatment or any outcome.
• We do not claim any regulatory clearance or approval for it.

As the product is set up today, the range beside the estimate is not a measured error bar. It comes from a fixed assumption about how much evidence sits behind an estimate. High, moderate and low precision are labels on the width of that range. None of those labels tests whether the estimate is right.

A language model, not a person, writes some of the explanatory text next to a probability, and nothing on screen marks it as machine written. Some disease incidence figures are model estimates too. Those are marked on screen, and are never used in a valuation.

Valuations and rNPV

The rNPV figures in the Finance tool are placeholders, as set out above. What follows is about valuations inside the Trial Terminal.

A risk adjusted net present value, or rNPV, is an arithmetic model. It takes assumptions about price, uptake, timing, cost and probability, and discounts them to a value today. The arithmetic is real. The assumptions are ours, and small changes to them move the answer a long way. Any figure we publish is one scenario. It is not a valuation of a company or an asset.

A valuation screen can be full of defaults. If nobody at Panoptic has entered assumptions for a trial, the screen still shows a complete sheet built from default values, including a placeholder probability of 50 percent. It will look like an analysis, and it is not one.

Catalyst dates

Regulatory dates, such as PDUFA dates and advisory committee meetings, come from a commercial calendar feed we subscribe to from BioPharmCatalyst. The Finance pages do not draw on that feed, and their catalyst dates are hardcoded, as set out above. We do not set the dates in the feed, we do not confirm them with the FDA, and we do not guarantee they are current. Dates move, so check any date against the company's own filings and the regulator before you rely on it.

Openness scores in the model registry

The Foundation Model Registry lists biological foundation models with an openness score. We do not calculate that score. We copy it from an outside registry that publishes it, along with the written model descriptions, the citations and the tag list. Our list is ordered by that score, so when we say models are ranked, we are repeating that registry's judgement rather than making our own.

Other preview screens

Some floors of the BioTower are design previews, and are marked as such on screen. Treat every figure on a preview screen as a placeholder.

What not to do with any of this

• Do not use our numbers as a basis for buying or selling any security. Nothing we publish is investment advice or a recommendation.
• Do not use them to make a clinical or treatment decision. Nothing we publish is medical advice.
• Do not treat them as regulatory, legal or tax advice.
• Do not rely on any figure without checking the underlying registry record and the company's own public filings.

If a number matters to a decision you are about to make, check it at the source first.

Questions

Email us at info@panoptic.bio.