Theranos: which warning signals a decision model scores early

Why does missing evidence count as a signal in the Theranos case?

The EQOS Asia Theranos page states that its data came from public sources available in 2013: a PubMed search, regulatory records, company materials, and professional networks (EQOS Asia Theranos case page, as of October 2026). The central finding is an absence. A search of the scientific literature for the company returned 0 results, after 10 years of work.

A decision model treats the absence of expected evidence as information. When a claim of scientific novelty exists, peer-reviewed work is the normal trace, so its absence lowers transparency and trust. The check takes minutes, and the signal arrives before any technical due diligence starts.

Which public checks does a decision model run first on Theranos?

The page presents signals that it says anyone could find in about 10 minutes (EQOS Asia Theranos case page, as of October 2026). A model can sequence such checks from the cheapest to the costliest: a literature search, a regulatory status review, a look at the board, then a request to see the technology working. Each step has a clear owner and a clear yes or no.

Sequencing protects time and budget. Each completed check updates a dimension score, and the running total shows whether the next, more expensive check is worth requesting. A reader who applies this to a supplier, a partner, or an acquisition target gets a clear rule: run the free checks first, then decide how much deeper to go. The sequence also documents the reasoning, so a colleague can follow why a decision waited.

Two companion pages extend this sequence. The Theranos red flags and timeline page places each signal on a calendar from 2013 to 2022 and shows which check fits each date. The page on what went wrong at Theranos reads the case in 4 layers, evidence, governance, openness, and patient safety, and turns each layer into a gate a team can apply before a commitment.

How does board expertise become a Theranos signal for a model?

The page compares the expertise of the board with the mission of the company, which was blood diagnostics, and finds a gap: the listed members came from diplomacy, the military, and politics, with no medical competence (EQOS Asia Theranos case page, as of October 2026). The model reads this as low alignment between competence and mission. The same check applies to any firm that asks for trust in a technical claim.

Governance signals work in every sector. A decision model asks who can challenge the technical claims, whether they are independent, and whether their expertise matches the work. The back-test scores integrity at 0.03, and its governance condition asks for at least 2 independent medical experts on the board.

What does secrecy about a product tell a decision model?

The page records the absence of a public demonstration of the device, broad non-disclosure agreements, and compartmentalised staff. It computes a ratio of 19 to 1 between notoriety and verifiable transparency, against a normal alert threshold below 3 to 1 (EQOS Asia Theranos case page, as of October 2026).

The lesson is about proportion. A company can protect secrets and still show evidence: independent testing, audited results, regulator filings. A model measures the gap between public visibility and verifiable transparency, and a wide gap raises the priority of independent verification. Notoriety is a poor substitute for evidence. Independent verification turns a claim into material that others can examine.

Why does the ethical safeguard block a Theranos-type decision?

The page documents patient cases where test results diverged from reality, and it scores care at 0.04 (EQOS Asia Theranos case page, as of October 2026). Under the EQOS Asia method, a negative ethical safeguard triggers an absolute block, whatever the other scores show. Here the block covers both investment and deployment.

A safeguard of this kind changes the shape of a decision. Financial scores answer whether a position pays, while the safeguard answers whether the position harms the people it claims to serve. A model that checks both gives a decision maker a documented reason to wait, even when growth figures look attractive. The check costs little to run, and it protects the reader while a claim awaits validation by an outside body.

Which four conditions would a Theranos-type model request?

The page lists four conditions: at least 1 peer-reviewed article in a leading journal, an independent laboratory review, a board with medical experts, and regulatory validation before patient use (EQOS Asia Theranos case page, as of October 2026). Each condition raises a named dimension, for example transparency from 0.02 to 0.40.

These conditions work as a template. A reader can copy the structure for another sector: one proof of performance, one independent technical review, one governance fix, and one external approval. The list stays short because a short list is easier to request, to verify, and to hold to account.

How can you use the Theranos lessons in a decision of your own?

Pick one claim that carries your decision, then list the evidence a sceptical expert would expect. The free simulator turns your ratings on 9 dimensions into a φ(S) estimate, and the score shows which dimension to examine first (EQOS Asia simulator page, as of October 2026).

For a structured read, the method page describes the Audit pillar and its 3 priority risks, and the case studies show what each tier delivers. The Theranos back-test is a retrospective exercise on public data, and it carries no investment advice or promise about future results.

Frequently asked questions

About the author

Stéphane Jambu

President, EQOS Forecast Asia

Stéphane Jambu leads EQOS Forecast Asia and works as a consultant in e-commerce, search visibility, and AI. LinkedIn profile