Level 1 - Absolute Beginner
Doctors at the Mayo Clinic made a computer program. It looks at medical pictures. It can find a sickness called pancreatic cancer early.
The program is called REDMOD. It looked at old scans that doctors said were normal. It found many signs that the doctors did not see.
Pancreatic cancer is very hard to find. Finding it early can help people a lot. The scientists say they need more tests before doctors use the program every day.
- doctor
- A person who helps sick people.
- program
- A set of computer instructions.
- scan
- A medical picture of the inside of the body.
- cancer
- A serious sickness where bad cells grow.
- pancreas
- A body part near the stomach.
- early
- Before the usual time.
- normal
- Not strange or sick.
- test
- A way to check if something works.
Level 2 - Elementary
Researchers at the Mayo Clinic and MD Anderson built an artificial intelligence model called REDMOD. It studies routine CT scans and looks for tiny signs of pancreatic cancer that doctors may miss.
The study, published in the journal Gut, tested nearly 2,000 CT scans that had first been read as normal. The model found about 73 percent of the cancers that were diagnosed later, on average around 16 months before diagnosis. Specialists reading the same scans without the AI found about 39 percent.
Pancreatic cancer is often found too late. Experts say these results are exciting, but the work looked back at old scans. Doctors still need tests on new patients in real time before hospitals can use REDMOD.
- artificial intelligence
- Computer systems that do tasks that need human thinking.
- routine
- Done regularly as a normal part of care.
- CT scan
- A detailed X-ray picture of the inside of the body.
- diagnose
- To find out what sickness a person has.
- specialist
- A doctor who focuses on one area.
- journal
- A magazine where scientists publish studies.
- patient
- A person who gets medical care.
- real time
- Happening now, not later.
Level 3 - Intermediate
Pancreatic cancer is among the deadliest cancers because it usually produces no obvious symptoms until it has spread. A study from the Mayo Clinic and MD Anderson, published in the journal Gut and in the news again this week, suggests artificial intelligence may help doctors spot it far earlier on scans they already take.
The model, called REDMOD, was tested on nearly 2,000 CT scans that radiologists had originally read as normal. It flagged about 73 percent of the cancers that were diagnosed later, with a median lead time of roughly 16 months, while specialists reading the same scans without AI flagged about 39 percent. On scans taken more than two years before diagnosis, it found nearly three times as many early cancers.
The authors and outside experts urge caution. The analysis was retrospective, meaning it looked back at images after the outcome was known, and the model has not yet been tested prospectively on new patients in routine care. Specificity was about 88 percent, so false alarms would also need to be managed before any screening use.
- symptom
- A sign of illness that a person feels or shows.
- radiologist
- A doctor who reads medical images.
- flag
- To mark something as needing attention.
- median
- The middle value in a set of numbers.
- lead time
- The time between an early warning and an event.
- retrospective
- Looking back at past data.
- prospectively
- Moving forward in time, testing on new cases.
- specificity
- How well a test avoids false alarms.
Level 4 - Advanced
Pancreatic cancer remains one of oncology's grimmest problems, in large part because it announces itself so late. A study by the Mayo Clinic and MD Anderson, published in the journal Gut and returning to the headlines this week, offers a glimmer of a different story: an artificial intelligence model, REDMOD, that picks out faint traces of malignancy on routine CT scans long before a diagnosis is made.
Tested on nearly 2,000 scans originally read as normal, the model identified about 73 percent of cancers that were diagnosed later, at a median lead time of roughly 16 months, against about 39 percent for specialists reading without it. On images taken more than two years before diagnosis, it detected nearly three times as many early tumours as the human readers did.
Yet the authors and independent specialists temper the enthusiasm. The work is retrospective, drawn from images whose outcomes were already known, and has yet to be validated prospectively in live clinical practice. With specificity of roughly 88 percent, the cost of false positives, in anxiety, further imaging and invasive procedures, will weigh heavily on whether the technique can ever be deployed for screening.
- oncology
- The branch of medicine that deals with cancer.
- glimmer
- A faint sign of hope.
- malignancy
- A cancerous growth.
- tumour
- An abnormal lump of cells.
- temper
- To soften or moderate.
- validate
- To confirm that something works as claimed.
- false positive
- A result that wrongly shows a problem.
- deploy
- To put into use.
Talk about this article with KatePractice this story's vocabulary in a real spoken conversation with your AI tutor