| The following article features coverage from the AACR Annual Meeting 2022. Click here to read more of Cancer Therapy Advisor’s conference coverage. |
An artificial intelligence (AI) model can use sequential health information from electronic health records (EHRs) to identify patients with a higher risk of developing pancreatic cancer, according to a poster presented at the AACR Annual Meeting 2022.
This type of model could enable earlier detection of pancreatic cancer and improve treatment options for patients, according to researchers.
“The purpose of this study was to develop an artificial intelligence tool that can help clinicians identify people at high risk for pancreatic cancer so they can be enrolled in prevention or surveillance programs and hopefully benefit from early treatment,” study author Bo Yuan, a PhD candidate at Harvard University in Boston, said in a statement.
Continue Reading
Yuan and colleagues used advanced machine learning technology to look at the time sequence of clinical events and predict the risk of cancer occurrence over a multi-year time interval.
The researchers used data from the Danish National Patient Registry, which includes EHRs for 8.6 million patients treated between 1977 and 2018. In this cohort, roughly 40,000 patients developed pancreatic cancer.
The researchers tested 4 machine learning models — 2 that were time-dependent (gated recurrent unit and transformer) and 2 that were not (bag of words and multilayer perceptron). The models were trained on the time sequence of diseases in patient clinical histories. The researchers tested the models’ ability to predict cancer occurrence in time intervals of 3 months to 60 months after risk assessment.
For cancer occurrence within 36 months, the best model — transformer — had an area under the receiver operating characteristic curve (AUROC) of 0.88. The odds ratio (OR) was 47.5 for 20% recall and 159.0 for 10% recall.
Yuan and colleagues validated their findings using EHR data from the Mass General Brigham Health Care System. Differences between the 2 datasets required the AI model to be retrained.
Even with the new training, the transformer model had comparable accuracy in both datasets. In the Mass General dataset, the AUROC was 0.87. The OR was 112.0 for 20% recall and 162.4 for 10% recall.
“These results indicate the potential of advanced computational technologies, such as AI and deep learning, to make increasingly accurate predictions based on each person’s health and disease history,” Yuan said.
Disclosures: This research was supported, in part, by the Novo Nordisk Foundation and Stand Up to Cancer. Please see the original reference for a full list of disclosures.
Read more of Cancer Therapy Advisor’s coverage of AACR 2022 by visiting the conference page.
Reference
Placido D, Yuan B, Hjaltelin JX, et al. AI predicts risk of pancreatic cancer from disease trajectories using real-world electronic health records (EHRs) from Denmark and the USA. Presented at AACR 2022; April 8-13, 2022. Abstract LB550.
