The same imaging model that finds a pancreatic tumor a year before your doctor can is running on the very capacity that also lets it read your self-reported race off a chest X-ray, on grounds no one can name. One fact, two sides. You cannot keep the first and discard the second.
The wins are real. A Mayo model flagged 73 in every 100 pre-diagnostic pancreatic cancers, against 39 for specialists reading the same scans unaided, a median of about sixteen months early. But found earlier is not the same as a life saved, and the same capacity underdiagnoses women, younger patients, and Black patients, Black women worst of all. An average is exactly the place a subgroup goes to hide.
