October 3, 2023

New AI Model Predicts Cancer Patient Survival More Accurately Than Previous Methods

” Predicting cancer survival is an essential factor that can be used to enhance cancer care,” said lead author Dr. John-Jose Nunez, a psychiatrist and scientific research study fellow with the UBC Mood Disorders Centre and BC Cancer. “These files have numerous details like the age of the patient, the type of cancer, underlying health conditions, past compound usage, and family histories. The scientists trained and tested the model using information from 47,625 clients across all six BC Cancer websites situated throughout British Columbia. To protect privacy, all client information remained kept safely at BC Cancer and was provided anonymously. In another stream of work, Dr. Nunez is analyzing how to facilitate the best-possible psychiatric and therapy care for cancer clients using innovative AI strategies.

Typically, cancer survival rates have actually been calculated retrospectively and categorized by just a couple of generic aspects such as cancer website and tissue type. In spite of familiarity with these rates, it can be challenging for oncologists to accurately forecast an individual clients survival due to the many intricate elements that influence client results.
The design developed by Dr. Nunez and his collaborators, which consists of scientists from BC Cancer and UBCs departments of computer science and psychiatry, has the ability to pick up on distinct clues within a patients preliminary consultation document to offer a more nuanced evaluation. It is likewise relevant to all cancers, whereas previous designs have actually been limited to certain cancer types.
” The AI basically checks out the assessment file similar to how a human would read it,” stated Dr. Nunez. “These documents have numerous details like the age of the client, the type of cancer, underlying health conditions, past compound usage, and family histories. The AI brings all of this together to paint a more total photo of patient results.”
The researchers trained and evaluated the model utilizing information from 47,625 clients throughout all six BC Cancer websites located throughout British Columbia. To protect personal privacy, all patient data stayed kept firmly at BC Cancer and was provided anonymously. Unlike chart evaluations by human research study assistants, the brand-new AI method has actually the added benefit of keeping complete privacy of patient records.
” Because the design is trained on B.C. data, that makes it a possibly powerful tool for anticipating cancer survival here in the province,” said Dr. Nunez.
In the future, the technology might be used in cancer clinics throughout Canada and around the world.
” The excellent aspect of neural NLP designs is that they are extremely scalable, portable, and dont need structured data sets,” stated Dr. Nunez. “We can quickly train these designs utilizing regional data to improve efficiency in a new region. I would believe that these models supply a great structure anywhere in the world where patients have the ability to see an oncologist.”
Dr. Nunez is a recipient of the 2022/23 UBC Institute of Mental Health Marshall Fellowship, and is likewise supported by moneying from the BC Cancer Foundation. In another stream of work, Dr. Nunez is taking a look at how to assist in the best-possible psychiatric and counseling care for cancer clients utilizing innovative AI methods. He pictures a future where AI is incorporated into many elements of the health system to enhance client care.
” I see AI acting nearly like a virtual assistant for doctors,” stated Dr. Nunez. “As medication gets more and more advanced, having AI to assist sort through and make sense of all the data will help notify physician choices. Ultimately, this will help enhance quality of life and results for patients.”
Reference: “Predicting the Survival of Patients With Cancer From Their Initial Oncology Consultation Document Using Natural Language Processing” by John-Jose Nunez, MD, MSc, Bonnie Leung, MN-NP( F), Cheryl Ho, MD, Alan T. Bates, MD, Ph.D. and Raymond T. Ng, Ph.D., 27 February 2023, JAMA Network Open.DOI: 10.1001/ jamanetworkopen.2023.0813.
The research study was funded by the BC Cancer Foundation.

Anticipating cancer patient survival rates is a vital element of cancer treatment and management. Properly anticipating a clients diagnosis helps physician make notified decisions about the most suitable course of action and can likewise aid in the development of tailored treatment plans.
Researchers from the University of British Columbia and BC Cancer have developed an AI design that forecasts cancer patient survival with greater precision and using more readily available information compared to previous methods.
This is the very first step in a cancer patients journey after diagnosis. The design was able to identify distinct features for each patient, resulting in survival predictions with over 80% precision for 6 months, 36 months, and 60 months.
” Predicting cancer survival is an essential factor that can be utilized to enhance cancer care,” stated lead author Dr. John-Jose Nunez, a psychiatrist and medical research fellow with the UBC Mood Disorders Centre and BC Cancer. “It may suggest health service providers make an earlier recommendation to support services or use a more aggressive treatment choice upfront. Our hope is that a tool like this might be utilized to individualize and enhance the care a patient gets right now, providing the very best result possible.”