Pancreatic cancer poses a formidable challenge due to its elusive nature in the early stages, often diagnosed when it has already advanced. To address this, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) collaborated with Dr. Limor Appelbaum from the Beth Israel Deaconess Medical Center to create an AI system that predicts the likelihood of developing pancreatic ductal adenocarcinoma (PDAC), the most prevalent form of pancreatic cancer. The study is published in eBioMedicine.
Named PRISM, the AI system outperformed existing diagnostic standards, offering a potential breakthrough in early detection. The system’s objective is to predict a patient’s risk of PDAC diagnosis within the next six to 18 months, enhancing the chances of early intervention.
PRISM comprises two AI models. The first utilizes artificial neural networks to identify patterns in electronic health records, considering factors like age, medical history, and lab results to calculate an individual risk score. The second model, using a simpler algorithm, generates a similar score. The models were trained on anonymized data from 6 million electronic health records, including 35,387 PDAC cases from 55 healthcare organizations in the US.
Results showed that the neural network identified 35% of individuals who developed pancreatic cancer as high risk six to 18 months before diagnosis, a significant improvement over current screening methods, which typically catch around 10% of cases.
PRISM’s potential impact on early detection is promising, according to Michael Goggins, a pancreatic cancer specialist at Johns Hopkins University School of Medicine. While the study is retrospective, the researchers are embarking on a prospective study, collecting data on existing patients to validate the accuracy of the predictions.
PRISM could revolutionize pancreatic cancer screening by pinpointing high-risk individuals for targeted testing and prompting broader screenings for asymptomatic individuals through blood or saliva tests. Dr. Appelbaum emphasizes the potential to save lives, noting the importance of transitioning such models from academic literature to clinical practice. As the researchers continue their work, PRISM offers hope for an effective tool in the early diagnosis of pancreatic cancer.