Artificial intelligence that scans GP records to find hidden patterns has significantly improved cancer detection rates. The use of the “C the Signs” AI tool has increased the detection rate from 58.7% to 66.0% in GP practices. This tool analyzes a patient’s medical record, including medical history, test results, prescriptions, treatments, and other personal characteristics such as postcode, age, and family history, to identify cancer risk.

The AI also prompts GPs to ask patients about new symptoms and recommends appropriate tests or clinical pathways if it detects patterns indicating a higher risk of cancer. “C the Signs” is used in about 1,400 practices in England, roughly 15%, and was tested in 35 practices in the east of England in May 2021, covering a population of 420,000 patients. Results published in the Journal of Clinical Oncology showed the cancer detection rate rose from 58.7% to 66.0% by March 31, 2022, while practices not using the system remained at a similar rate.

Bea Bakshi, a GP who created the system with colleague Miles Payling, stated, “It could be a scan, an ultrasound, or they could need to be seen by a specialist at a clinic.” The system tracks patients to remind doctors to check test results and referrals. “Our system has detected over 50 different types of cancers,” Bakshi added. “The key thing is that it’s not only an earlier diagnosis but a faster diagnosis.”

In a previous study, the tool assessed 118,677 patients, diagnosing 7,295 with cancer, with the algorithm successfully identifying 7,056. Only 239 out of 8,453 patients initially deemed unlikely to have cancer were confirmed to have it within six months, about 2.8%. Bakshi developed the tool after a patient she met in hospital received a late diagnosis of pancreatic cancer and died three weeks later. This experience highlighted the problem of late cancer diagnoses.

The UK has screening programs for bowel, breast, and cervical cancers, but there are about 200 different types of cancer that may be asymptomatic or have symptoms easily confused with other conditions. “Two-thirds of deaths are in the non-screenable cancers,” Bakshi said. “Patients visit GPs three to five times before being recognized as at risk of cancer. GPs detect an average of eight cases of cancer a year.”

GPs use guidelines from the National Institute for Health and Care Excellence for cancer referrals. “They’re comprehensive guidelines, but no GP can remember them all,” said Peter Holloway, a GP and co-author of the study. “Many cancers present with vague symptoms that don’t necessarily align with our guidelines.”

Holloway shared an example of a patient in his early 60s with common symptoms of diarrhea and lower abdominal pain. The C the Signs tool recommended a faecal test, which was positive, leading to an early and successful diagnosis of colorectal cancer. “If we’d followed the rigid guidelines, he may not have been referred for several months,” Holloway said.

The NHS England Long Term Plan for Cancer aims to diagnose 75% of all cancers at stage one or stage two by 2028. The NHS is also researching the effectiveness of the Galleri blood test, which attempts to detect DNA from more than 50 different types of tumors. The trial, which began in September 2021, has tested 140,000 people.

Peter Holloway emphasized that decision support systems like C the Signs are crucial for cancer detection. He also highlighted the importance of improving patient awareness of cancer symptoms and enhancing access to diagnostic technology such as CT and MRI scanners.

Professor Peter Johnson, National Clinical Director for Cancer at NHS England, stated, “Despite increased demand and pressure on services, record numbers of people are being checked and treated for cancer. We are now diagnosing a higher proportion of cancers at an early stage, which increases survival rates.

“We know there is much more to do to ensure people with cancer receive the care they need. Using the latest technology is a key part of our efforts to reduce waiting times and detect cancers earlier. This includes ‘teledermatology’ for diagnosing skin cancers, community lung trucks, and home testing for bowel cancer.”