erictopol.substack.com/p/dawn-of-a-new-era-of-primary-prevention
3 corrections found
which is essentially the complete list of International Classification of Diseases (ICD-10)
This overstates the paper. The Nature study modeled more than 1,000 diagnoses at the top level of ICD-10, not the complete ICD-10 classification.
Full reasoning
The article conflates top-level ICD-10 diagnoses with the entire ICD-10 system.
In the Nature paper being discussed, the authors say their model predicts "more than 1,000 diseases" and specifically explain that these are diagnoses "at the top level of the International Classification of Diseases, Tenth Revision (ICD-10) coding system." Later in the paper they state that the model vocabulary includes "ICD-10 top-level diagnostic codes."
WHO’s own ICD-10 browser documentation also makes clear that ICD-10 is hierarchical: it says the browser initially shows only the top level items, and users can expand those items to reveal child categories underneath. So the study did not cover the complete ICD-10 classification; it used the top-level layer of ICD-10 diagnoses.
2 sources
- Learning the natural history of human disease with generative transformers
few algorithms are capable of predicting the full spectrum of human disease, which recognizes more than 1,000 diagnoses at the top level of the International Classification of Diseases, Tenth Revision (ICD-10) coding system... The vocabulary of the model includes ICD-10 top-level diagnostic codes
- WHO ICD-10 User Guide
When browsing ICD10, you will see the classification hierarchy... Initially, the system only shows the top level items. However, you may make the children visible by clicking on the small triangles on the left side of the items.
thus far no disease-modifying impact, for neurodegenerative diseases (Alzheimer’s and Parkinson’s).
This is incorrect because Alzheimer’s disease now has FDA-approved disease-modifying treatments that have verified clinical benefit and slow clinical decline.
Full reasoning
The statement is too broad because it explicitly includes Alzheimer’s disease, for which the FDA has already recognized disease-modifying clinical benefit.
- In July 2023, the FDA converted Leqembi (lecanemab) to traditional approval after a confirmatory trial "verified clinical benefit." The FDA said this action was "the first verification that a drug targeting the underlying disease process of Alzheimer's disease has shown clinical benefit" and reported a "statistically significant and clinically meaningful reduction of decline" versus placebo.
- In July 2024, the FDA approved Kisunla (donanemab) and stated that treated patients "demonstrated a statistically significant reduction in clinical decline" versus placebo.
That does not mean Alzheimer’s has been cured, nor that Parkinson’s has an equivalent approved therapy. But the article’s categorical claim that there has been "no disease-modifying impact" for neurodegenerative diseases including Alzheimer’s is contradicted by current FDA determinations and trial results for approved Alzheimer’s treatments.
2 sources
- FDA Converts Novel Alzheimer's Disease Treatment to Traditional Approval
Today, the U.S. Food and Drug Administration converted Leqembi ... to traditional approval following a determination that a confirmatory trial verified clinical benefit... 'the first verification that a drug targeting the underlying disease process of Alzheimer's disease has shown clinical benefit' ... Leqembi demonstrated a statistically significant and clinically meaningful reduction of decline.
- FDA approves treatment for adults with Alzheimer's disease
Patients treated with Kisunla demonstrated a statistically significant reduction in clinical decline on the Integrated Alzheimer's Disease Rating Scale (iADRS) compared to placebo at Week 76 in the overall population.
in over 90% of cases, is only currently diagnosed at late, metastatic stages.
This substantially overstates how often pancreatic cancer is metastatic at diagnosis. U.S. SEER data show about 51% of cases are distant/metastatic at diagnosis, not over 90%.
Full reasoning
The claim says pancreatic cancer is diagnosed at "late, metastatic stages" in over 90% of cases. That is not what current U.S. cancer surveillance data show.
According to the National Cancer Institute’s SEER Cancer Stat Facts for pancreatic cancer, the distribution at diagnosis is approximately:
- 15% localized
- 28% regional
- 51% distant (metastatic)
- 6% unknown
So pancreatic cancer is indeed often found late overall, but the share diagnosed as metastatic/distant is about half, not over 90%. The American Cancer Society likewise describes SEER staging for pancreatic cancer as localized, regional, and distant, with distant disease as only one of those categories.
A more accurate formulation would be that most pancreatic cancers are diagnosed after they have already spread beyond the pancreas, but not that over 90% are metastatic at diagnosis.
2 sources
- Pancreatic Cancer — Cancer Stat Facts
Percent of Cases & 5-Year Relative Survival by Stage at Diagnosis: Pancreatic Cancer ... Localized 15% ... Regional 28% ... Distant 51% ... Unknown 6%.
- Survival Rates for Pancreatic Cancer
The SEER database ... groups cancers into localized, regional, and distant stages: Localized ... Regional ... Distant: The cancer has spread to distant parts of the body, such as the lungs, liver, or bones.