“There is a lot of useful information in a medical record, both over time and across different disease areas,” said Giovanni Parmigiani, a Dana-Farber researcher who is one of the senior authors of the paper. “That data would be difficult for a human to process in their head, but [is manageable] for a machine learning model.”
In the past few years, scientists have experimented with using AI to sift through electronic health records to identify undiagnosed rare diseases or to detect the earliest signs of Alzheimer’s.
The tool is called Aladynoulli. That’s a portmanteau blending Aladdin of folk tales and Disney movie fame for his magical genie-summoning powers, and Jacob Bernoulli, the 17th-century Swiss mathematician who helped pioneer the subject of probability.
The researchers hope Dana-Farber and MGH incorporate the AI tool into their electronic health records in the near future but couldn’t say when. Nor did they know when it could become available to patients elsewhere. Parmigiani said the technology is “ready for prime time” but added that AI tools face a complex journey to reach widespread use.
To develop the AI model and increase its reliability, Aladynoulli’s creators used data contained in 683,000 patient records. Many of the records came from the UK Biobank, which houses health-related data from consenting patients without identifying them and is headquartered in Greater Manchester, England.
The researchers used the algorithm to home in on 20 biological trends that increase the risk of causing certain diseases to occur. There are, for example, 17 conditions in a cluster associated with colorectal cancer risk, including ulcerative colitis, general gastrointestinal complications, and hemorrhoids, Parmigiani said.
Aladynoulli is particularly impressive in predicting which patients will develop colorectal cancer in the coming year based on findings in electronic health records, the investigators said.
If the tool identifies such a patient, a physician could refer that individual for a colonoscopy, even if the patient isn’t scheduled for the diagnostic test yet on the basis of age-related guidelines.
If the colonoscopy indicates that patient may be developing colorectal cancer or polyps that could be a precursor, a doctor could intervene early, when the disease is still treatable or preventable.
There has been a sharp increase in colorectal cancer in younger adults in the United States in recent years. It is now the leading cause of cancer death in adults under 50, according to Emory Healthcare.
Some doctors already use online calculators to measure the risk of patients developing certain diseases, including PCE, QRISK3, and PREVENT for cardiovascular disease and GAIL for breast cancer. The researchers who wrote the Nature paper said their algorithm outperformed them all and measured the risk of many more diseases.
Not surprisingly, said Dr. Sarah Urbut, an MGH cardiologist and lead author of the Nature paper, the tool gets better at predicting diseases as a patient sees more doctors and generates more electronic health records.
“It sees the patient holistically, both across departments and over time,” said Urbut, who has worked on the algorithm for about three years.
She said, for example, a rheumatologist treating a patient for an inflammatory disease could make notes that might be relevant to a cardiologist, given links between some inflammatory conditions and risks for developing heart disease.
“Physicians are extremely conscientious, and we do often read notes from many other specialties,” she said. “But an interview is limited, and if we had an algorithm that said which notes we should pay attention to and what we could learn from other departments, that’s amazing.”
Jonathan Saltzman can be reached at jonathan.saltzman@globe.com.