Researchers from the University of British Columbia and BC Cancer have created an AI model that predicts cancer patient survival with greater accuracy using natural language processing (NLP). This AI model examines oncologists’ notes taken following a patient’s initial consultation and is able to identify distinctive features for each patient, resulting in survival predictions with over 80% accuracy for 6 months, 36 months, and 60 months. This is the first AI model of its kind to use NLP to predict cancer patient survival and its findings were recently published in the JAMA Network Open.
UBC researchers have developed a new tool to predict cancer survival rates. The tool, which uses machine learning algorithms, can help health providers personalize and optimize care for cancer patients. Lead author Dr. John-Jose Nunez, a psychiatrist and clinical research fellow with the UBC Mood Disorders Centre and BC Cancer, believes this tool could be used to make earlier referrals to support services or offer more aggressive treatment options upfront. By predicting cancer survival, this tool could help improve cancer care and give patients the best outcome possible.
Dr. Nunez and his team of researchers from BC Cancer, UBC’s departments of computer science and psychiatry have developed a model that can accurately predict an individual patient’s cancer survival rate. This model is based on unique clues found within a patient’s initial consultation document and is applicable to all cancer types, unlike previous models which were limited to certain cancer types. This model provides a more nuanced assessment of a patient’s survival rate and can help oncologists make more informed decisions about treatment. It is a valuable tool for improving patient outcomes and providing more personalized care.
The AI reads consultation documents, such as age, type of cancer, underlying health conditions, past substance use, and family histories, to paint a more complete picture of patient outcomes. The AI was trained and tested using data from 47,625 patients across all six BC Cancer sites located across British Columbia. The AI approach has the added benefit of maintaining complete confidentiality of patient records, ensuring patient privacy is protected. This AI system can help healthcare providers make more informed decisions about patient care and treatment.
This new developed powerful tool to predict cancer survival rates are using natural language processing (NLP) models. This technology is trained on data from British Columbia, making it a powerful tool for predicting cancer survival in the province. The great thing about neural NLP models is that they are highly scalable, portable, and don’t require structured data sets, meaning they can be quickly trained using local data to improve performance in a new region. This technology could be applied in cancer clinics across Canada and around the world, providing a good foundation for predicting cancer survival anywhere patients are able to see an oncologist.
Dr. Nunez is a recipient of the 2022/23 UBC Institute of Mental Health Marshall Fellowship and is supported by the BC Cancer Foundation. He is researching how to use advanced AI techniques to provide the best possible psychiatric and counseling care for cancer patients. Dr. Nunez envisions a future where AI is integrated into the health system to improve patient care, acting as a virtual assistant for physicians to help make sense of data and inform decisions. This could lead to improved quality of life and outcomes for patients.
