Texas AI Docket

An MD Anderson and UT Medical Branch MRI model scored lower on patients it was not built on

Health and educationThe University of Texas MD Anderson Cancer Center, with The University of Texas Medical BranchHarris, GalvestonWrite to the decider

The authors include radiologists at the University of Texas MD Anderson Cancer Center in Houston and at the University of Texas Medical Branch in Galveston. They built a machine learning survival model from MRI radiomic features and tested it on a separate patient group without refitting it. Its discrimination score was lower on the group it was tested on than on the group it was built on. The paper reports overlapping confidence intervals for the two and no test that the difference is real. The authors conclude that the approach showed limited standalone discrimination and support cautious use of it as an exploratory imaging biomarker. The paper is a research result rather than a deployment, and neither institution has published a statement that the model is used in patient care.

How to take part

The paper is published in the American Journal of Neuroradiology. The abstract, the figures it reports and the author affiliations are readable without a subscription through the National Library of Medicine.

Where to do it

Where

2 counties. In Houston-Pasadena-The Woodlands.

Timeline

  1. filed

    Article date recorded by the National Library of Medicine for the online publication

  2. Today

How this decision moved

One dated line per check, oldest first. A line that says nothing changed means somebody looked and it had not.

  1. 2026-09-13

    Admitted to the record. A model built at a Texas cancer center scored lower on patients it had not been fitted to. The authors call that limited standalone discrimination and ask for cautious use, and they report no test that the difference is real.

  2. 2026-09-16

    The MRI model still scored lower on patients it was not built on, and the authors' own reading of that result is unchanged. Nothing has been retracted.

  3. 2026-09-19

    The result stands as published, and the model has not been reported as refitted to the patients it scored lower on.

  4. 2026-09-23

    The model still scored lower on the patients it was not built on, which is what the paper reported.

  5. 2026-09-26

    The paper still reports the model scoring lower on the patients it was not built on.

The evidence

Every fact above rests on one of these. The words are the source's own.

This retrospective study developed a T1 postcontrast MRI radiomics survival model in a public brain metastasis cohort of 198 patients and externally validated the fixed radiomics model, without refitting or recalibration, in an independent cohort of 69 patients.
MRI Radiomics for Survival Prediction in Brain Metastases: A Machine Learning Analysis, National Library of Medicine Primary source, official · eutils.ncbi.nlm.nih.gov
The radiomics score had a C-index of 0.615 (95% CI, 0.543-0.687) in the training cohort and 0.574 (95% CI, 0.491-0.658) in external validation.
MRI Radiomics for Survival Prediction in Brain Metastases: A Machine Learning Analysis, National Library of Medicine Primary source, official · eutils.ncbi.nlm.nih.gov
T1 postcontrast MRI radiomics showed limited standalone discrimination for overall survival in patients with brain metastases.
MRI Radiomics for Survival Prediction in Brain Metastases: A Machine Learning Analysis, National Library of Medicine Primary source, official · eutils.ncbi.nlm.nih.gov
These results support cautious use of radiomics as an exploratory imaging biomarker and emphasize the need for integrated prognostic models that include clinical, treatment, molecular, and systemic disease variables.
MRI Radiomics for Survival Prediction in Brain Metastases: A Machine Learning Analysis, National Library of Medicine Primary source, official · eutils.ncbi.nlm.nih.gov
From the Department of Neuroradiology (R.E., H.A.Q., S.A., A.M., P.K.), The University of Texas MD Anderson Cancer Center, Houston, TX 77030, USA; Department of Radiology (E.C.), Duke University Medical Center, Durham, NC, USA; Department of Radiology (H.A.S., A.N., M.W.), The University of Texas Medical Branch, Galveston, TX, USA
MRI Radiomics for Survival Prediction in Brain Metastases: A Machine Learning Analysis, National Library of Medicine Primary source, official · eutils.ncbi.nlm.nih.gov
The final model retained 7 nonzero T1 postcontrast radiomics features after correlation filtering and elastic-net selection.
MRI Radiomics for Survival Prediction in Brain Metastases: A Machine Learning Analysis, National Library of Medicine Primary source, official · eutils.ncbi.nlm.nih.gov
In the external-validation subset with available Graded Prognostic Assessment, the combined radiomics plus Graded Prognostic Assessment model had a C-index of 0.591 (95% CI, 0.509-0.672).
MRI Radiomics for Survival Prediction in Brain Metastases: A Machine Learning Analysis, National Library of Medicine Primary source, official · eutils.ncbi.nlm.nih.gov

Questions about this decision

Answered from the record itself. Every answer is assembled from stored fields, so an answer the record has no basis for is left out rather than guessed.

What is this decision?

The authors include radiologists at the University of Texas MD Anderson Cancer Center in Houston and at the University of Texas Medical Branch in Galveston. They built a machine learning survival model from MRI radiomic features and tested it on a separate patient group without refitting it. Its discrimination score was lower on the group it was tested on than on the group it was built on. The paper reports overlapping confidence intervals for the two and no test that the difference is real. The authors conclude that the approach showed limited standalone discrimination and support cautious use of it as an exploratory imaging biomarker. The paper is a research result rather than a deployment, and neither institution has published a statement that the model is used in patient care.

Who decides it?

The University of Texas MD Anderson Cancer Center, with The University of Texas Medical Branch decides. The record names the deciding body for every entry it carries.

Can the public take part?

The paper is published in the American Journal of Neuroradiology. The abstract, the figures it reports and the author affiliations are readable without a subscription through the National Library of Medicine. No dated public window is on the record. The deciding body is named and reachable.

Where in Texas does it apply?

It covers Galveston and Harris Counties in the Houston-Pasadena-The Woodlands, TX area.

Has it been decided?

It has been decided. The dates on the item page carry when.

What happens next?

No future date is on the record. The last dated step on it was filed on September 9th.

When did it start?

The earliest date on its record is September 9th, 2026.

What kind of decision is it?

It is filed under health and education.

What sources back it?

One source backs it. It is primary.

Is it on the ERCOT grid?

Yes. It sits inside the ERCOT interconnection.

When was it last checked?

Every fact on it was last verified against its source on September 26th, 2026.

Cite this

Texas AI Docket, An MD Anderson and UT Medical Branch MRI model scored lower on patients it was not built on. Tracked since September 9th, 2026. Last verified September 26th, 2026. https://texasaidocket.com/item/tx-2026-0151/. Reuse permitted under CC BY 4.0 with attribution. The same entry is in the docket JSON as item tx-2026-0151.

Beat

Filed under Health and education, with every other decision on that beat.

Last checked 2026-09-26