Claim

In a randomized controlled trial of 249 physicians in Indonesia, Kenya, and the Netherlands, GPT-4o access improved clinical-performance scores relative to control by 18 percentage points in Kenya, 10.7 in Indonesia, and 7.2 in the Netherlands under controlled conditions; harms were not assessed.

Type QUANTITATIVE · truth-evaluable: yes · status at snapshot: ACTIVE · created seq 17938

43 words; atomic claims are usually under 25a conjunction joins clauses that may be separately evaluablea comma-separated series may list several propositions

Referenced 116 times: checked 2, looked up by assistants 0, viewed 82, shared 32. A count of attention, not evidence.

Topics

AI · 1efd24fe Medicine · 1efd24fe

Assessment

Supported

Score: 0.9468 under ledger-default@0.3.0 at snapshot 27745

2 of 4 review checks done · 1 counted · 1 by the author only

2 independent evidence lineages · review checks 1 of 4 · stability HIGH

Main unresolved issue: 4 counted evidence items have no independence review; repeats of one origin may be counted as several coverage:eab632b1-2fce-8c07-9269-2130ece240d6

Not yet independently audited.Evidence reviewed so far supports this claim, but only 1 of 4 review checks has been done.A quoted passage from nature.com was not found on the page when Galedra fetched it.

Why?

Strongest counted support: The multicountry randomized trial reported country-specific performance gains with GPT-4o access but did not assess harms. weight 1.800000 evidence

Suppressed as dependent: The packet directly supports the sample, randomized GPT-4o comparison, country-specific performance gains, and unassessed harms.

Review gaps: qualifiers reviewed, opposing search done, independence reviewed

What would most change this: One independent, direct measurement that contradicts this claim would move it from Supported to Leans supported.

Hide calculation

0.9468 under ledger-default@0.3.0 at snapshot 27745 · state Supported · Review checks: 1 of 4 · stability HIGH · 4 sources, 2 independent lineages

Model ledger-default@0.3.0 (the default model, not the answer) · snapshot 27745 · trace sha256:ec555afa57389fd44e9d497fb54cd003a3fefe01ddcb5986eaadea9dd15dc2c1

Last changed at entry seq 25376; nothing logged since has borne on this claim.

How the number was reached

Each counted link is weighed by multiplying the factors the model declares; a factor it does not declare is not shown. Every factor name below links to what that number means.

Link	Working	Weight	Counted as
8384fdaf support	relevance DIRECT 2.0 × observation DATASET_RESULT 0.9 × interpretation 0 steps, 1 − 0.15 × 0 1.000000 × authenticity UNVERIFIED 1.0 × extraction UNVERIFIED 1.0 × provenance INDEPENDENT 1.0 × edition NAMED_OR_UNRELATED 1	1.800000	0.000000 (dependent strongest only, kept f39b55ea)
03af7712 support	relevance STRONG 1.2 × observation DIRECT_TEXT 0.9 × interpretation 0 steps, 1 − 0.15 × 0 1.000000 × authenticity UNVERIFIED 1.0 × extraction UNVERIFIED 1.0 × provenance INDEPENDENT 1.0 × edition NAMED_OR_UNRELATED 1	1.080000	1.080000
ef758388 qualify	relevance DIRECT 2.0 × observation DIRECT_TEXT 0.9 × interpretation 0 steps, 1 − 0.15 × 0 1.000000 × authenticity UNVERIFIED 1.0 × extraction UNVERIFIED 1.0 × provenance INDEPENDENT 1.0 × edition NAMED_OR_UNRELATED 1	1.800000	0.000000 (non directional)
f39b55ea support	relevance DIRECT 2.0 × observation DATASET_RESULT 0.9 × interpretation 0 steps, 1 − 0.15 × 0 1.000000 × authenticity UNVERIFIED 1.0 × extraction UNVERIFIED 1.0 × provenance INDEPENDENT 1.0 × edition NAMED_OR_UNRELATED 1	1.800000	1.800000

prior 0.50 → log-odds 0.000000 + 2.880000 (the counted weights) = 2.880000 → probability 0.9468

Under neighbouring settings of the same model: 0.8825 to 0.9769, a spread of 0.0944.

Half-even rounding, six places for weights and four for the probability. The same inputs under the same model reproduce this exactly; that is what the trace hash above is for. How a score is worked out explains every factor, why it is there, and what the model sets it to.

Your view

Registered views

Nobody has registered a view yet.

Registered views are personal belief, not evidence: they never change Galedra's assessment. Affiliations are self-declared and shown only for groups of 5 or more.

Checked as part of

Statements this claim was broken out of. The claim is scored on its own; a statement's reading only counts its claims.

“In a randomized controlled trial of 249 physicians in Indonesia, Kenya, and the Netherlands, GPT-4o access improved clinical-performance scores relative to...” · 1 claim · 29 September 2026
“In a randomized controlled trial of 249 physicians in Indonesia, Kenya, and the Netherlands, GPT-4o access improved clinical-performance scores relative to...” · 1 claim · 29 September 2026

Evidence for

The multicountry randomized trial reported country-specific performance gains with GPT-4o access but did not assess harms.

From Impact of LLM assistance on physician decision-making: a multi-country randomized controlled trial Read 2026-09-29 Relevance DIRECT Steps 0 Counted
The packet directly supports the sample, randomized GPT-4o comparison, country-specific performance gains, and unassessed harms.

From Impact of LLM assistance on physician decision-making: a multi-country randomized controlled trial Read 2026-09-29 Relevance DIRECT Steps 0 Counted
Independent ILO analysis corroborates the 249-physician randomized design and country-specific GPT-4o performance gains.

From Does a General-Purpose Large Language Model Improve Physicians’ Clinical Reasoning? Evidence and considerations for policy from a randomized trial in Indonesia, Kenya and the Netherlands Read 2026-09-29 Relevance STRONG Steps 0 Counted

Independent policy analysis reports the same trial’s principal results and controlled-vignette limitation.

Evidence against

None.

Qualifications

The country-specific score gains came from a controlled comparison where controls lacked internet and guidelines, and the study did not assess harms or real-practice outcomes.

From Impact of LLM assistance on physician decision-making: a multi-country randomized controlled trial Read 2026-09-29 Relevance DIRECT Steps 0 Counted

Clarifies comparator resources and limits generalization beyond controlled conditions.

Suppressed as dependent

Evidence	Group	Kept instead
e9f1f534		f39b55ea

Inferences

Concluded from

None.

Used as a premise in

None.

An inference is interpretation, not evidence: it never changes the assessment of its conclusion. Each premise is a claim with its own evidence.

Threads

None. Threads are for disagreements about how a determination was made — a statement whose figures are not in the passage it rests on, two sources that may share an origin. Something about the world belongs in the evidence above.

Claim edges

None. Edges are stored and displayed; v0.1 does not propagate probability across them.

Review checklist

Review checks: 1 of 4

2 checks by this claim's own author, alongside the independent review above. 1 QUALIFIER CHECK · 1 EVIDENCE VERIFICATION. A check by the person who recorded a claim does not raise its review coverage.

Check	Done	By
qualifiers reviewed	no
opposing search done	no
independence reviewed	no
primary source reviewed	yes	8384fdaf, ef758388, f39b55ea

Summary

Supported. The main support is: Independent ILO analysis corroborates the 249-physician randomized design and country-specific GPT-4o performance gains. 7059d71b

1 further item repeats the same origin and is not independent evidence. e9f1f534

Qualifier: The country-specific score gains came from a controlled comparison where controls lacked internet and guidelines, and the study did not assess harms or real-practice outcomes. 3c8051c7

Review checks: 1 of 4. coverage:eab632b1-2fce-8c07-9269-2130ece240d6

Generated by stub-v0.1; every sentence cites graph ids.

Contribution history

seq	event	detail
17938	APPENDED
17939	ACCEPT	AUTOMATIC_AFTER_VALIDATION
17944	LINK_EVIDENCE	by 1efd24fe (accepted)
17989	TASK_RESULT	by 1efd24fe (accepted)
17997	LINK_EVIDENCE	by 1efd24fe (accepted)
25375	TASK_RESULT	by 1efd24fe (accepted)

Score trace

{
  "claim": "eab632b1-2fce-8c07-9269-2130ece240d6",
  "links": [
    {
      "kept": "62f03340-d58c-82c8-80ab-03aff39b55ea",
      "link": "140ba635-2e0e-8e30-a8e9-a1938384fdaf",
      "group": null,
      "reason": "dependent_strongest_only",
      "edition": "NAMED_OR_UNRELATED",
      "evidence": "294e1e73-a42f-8693-8caf-c7c1e9f1f534",
      "direction": "SUPPORT",
      "magnitude": "1.800000",
      "relevance": "DIRECT",
      "extraction": "UNVERIFIED",
      "provenance": "INDEPENDENT",
      "observation": "DATASET_RESULT",
      "source_type": "PRIMARY_TEXT",
      "authenticity": "UNVERIFIED",
      "audit_confirmed": false,
      "effective_weight": "0.000000",
      "interpretive_steps": 0
    },
    {
      "link": "1fc61b43-d27b-8294-adf3-c31b03af7712",
      "group": null,
      "edition": "NAMED_OR_UNRELATED",
      "evidence": "e7893d0f-7ffb-803f-8309-3c867059d71b",
      "direction": "SUPPORT",
      "magnitude": "1.080000",
      "relevance": "STRONG",
      "extraction": "UNVERIFIED",
      "provenance": "INDEPENDENT",
      "observation": "DIRECT_TEXT",
      "source_type": "SECONDARY_TEXT",
      "authenticity": "UNVERIFIED",
      "audit_confirmed": false,
      "effective_weight": "1.080000",
      "interpretive_steps": 0
    },
    {
      "link": "311d192d-66ee-8603-9497-e8faef758388",
      "group": null,
      "reason": "non_directional",
      "edition": "NAMED_OR_UNRELATED",
      "evidence": "36e89840-7e97-8bbd-a80d-42603c8051c7",
      "direction": "QUALIFY",
      "magnitude": "1.800000",
      "relevance": "DIRECT",
      "extraction": "UNVERIFIED",
      "provenance": "INDEPENDENT",
      "observation": "DIRECT_TEXT",
      "source_type": "PRIMARY_TEXT",
      "authenticity": "UNVERIFIED",
      "audit_confirmed": false,
      "effective_weight": "0.000000",
      "interpretive_steps": 0
    },
    {
      "link": "62f03340-d58c-82c8-80ab-03aff39b55ea",
      "group": null,
      "edition": "NAMED_OR_UNRELATED",
      "evidence": "8f02c786-4001-87b2-8b9f-d076a9fb298a",
      "direction": "SUPPORT",
      "magnitude": "1.800000",
      "relevance": "DIRECT",
      "extraction": "UNVERIFIED",
      "provenance": "INDEPENDENT",
      "observation": "DATASET_RESULT",
      "source_type": "PRIMARY_TEXT",
      "authenticity": "UNVERIFIED",
      "audit_confirmed": false,
      "effective_weight": "1.800000",
      "interpretive_steps": 0
    }
  ],
  "model": "ledger-default@0.3.0",
  "prior": "0.50",
  "variants": {
    "spread": "0.0944",
    "permissive": "0.9769",
    "conservative": "0.8825"
  },
  "code_hash": "sha256:c61a6fdf35a754a0d7e958df149036a083c0eaeb7d2d2e52704ec32b0c36def7",
  "contested": false,
  "stability": "HIGH",
  "claim_type": "QUANTITATIVE",
  "config_hash": "sha256:86768038080b8ab5233f1e4b7aba10af05b5bb44caa432095150b90e81209809",
  "probability": "0.9468",
  "provisional": true,
  "evidence_sum": "2.880000",
  "snapshot_seq": 25376,
  "prior_log_odds": "0.000000",
  "support_groups": 2,
  "model_dependent": false,
  "review_coverage": "0.25",
  "truth_evaluable": true,
  "assessment_state": "SUPPORTED",
  "review_checklist": {
    "qualifiers_reviewed": {
      "by": [],
      "ok": false
    },
    "opposing_search_done": {
      "by": [],
      "ok": false
    },
    "independence_reviewed": {
      "by": [],
      "ok": false
    },
    "primary_source_reviewed": {
      "by": [
        "140ba635-2e0e-8e30-a8e9-a1938384fdaf",
        "311d192d-66ee-8603-9497-e8faef758388",
        "62f03340-d58c-82c8-80ab-03aff39b55ea"
      ],
      "ok": true
    }
  },
  "contradict_groups": 0,
  "rounding_boundary": false,
  "posterior_log_odds": "2.880000",
  "not_applicable_reason": null,
  "independence_unreviewed": 4
}