TRV-2026-2219Certified recordJournalism

Can AI really predict cancer survival better than a doctor? I have my doubts | Devi Sridhar

The idea that a single selfie could predict how long someone with cancer may live sounds like an incredible breakthrough. AI is getting better at spotting patterns in medical data, but can it be used to reliably predict what will happen to an individual patient? I asked a clinical colleague what she thought about the recent headlines. She found it an interesting development given that doctors have long used facial appearance to assess patient health. For example, she noted that they may write “appears stated age…

Health · The Trace — both readings · certified 2026-10-23 · v1 · article view · machine-readable

Current reading — gain

An AI model trained on facial images found AI-estimated facial age significantly associated with survival in 18 of 28 cancer categories, suggesting it could add prognostic information beyond a doctor's visual impression.

Current reading — problem

The facial-age measure was not significantly associated with survival for Black patients after adjustment, and selfie variability, cosmetic changes and retrospective design limit reliable individual prediction.

What this doesn’t fix

The tool has not been tested against doctors' own judgments or across different hospitals and patient groups, and the models rely on retrospective associations rather than proven clinical usefulness.

Evidence

Reader signal

How should this claim be treated?

Cite this record

Truvace Impact Record TRV-2026-2219, v1: “Can AI really predict cancer survival better than a doctor? I have my doubts | Devi Sridhar.” Truvace, 2026-10-23. /record/TRV-2026-2219 (accessed at citation time). sha256 6b916e6e6132bb91…

Calibration history

Every change to this record since certification, in the open. None yet — the reading has held since it entered the record.

  1. Certifiedv16b916e6e6132…

    Certified into the record

Verify this record
How to verify without trusting this page

Fetch the canonical text of any version from /api/record/TRV-2026-2219 and hash it yourself — for example shasum -a 256 on the saved canonical field. The result must equal content_hash, and each version’s text ends with prev:followed by the prior version’s hash (version 1 chains to 64 zeros). If a single character of any version had been altered since certification, the chain would not reproduce.