Scientists from the University of British Columbia have shown that artificial intelligence can accurately predict the survival rate of cancer patients by analyzing physician records. Research published JAMA Network Open.
The model developed by the scientists uses natural language processing (NLP). AI reads doctor’s notes as if they were read by a human. In doing so, the algorithm takes into account many details: the patient’s age, type of cancer, underlying diseases, past substance use, and family history. All these AI metrics combine to paint a more complete picture of patient outcomes.
The researchers trained and tested the model using data from 47,625 patients in British Columbia. The AI was able to predict survival rates at 6 months, 36 months, and 60 months with over 80% accuracy. While the model is applicable to all cancer types, previous algorithms were limited to certain tumor types.
Because of the many complex factors that affect patient outcomes, it is difficult for doctors to predict the survival of each patient. Accurately estimating the life expectancy of cancer patients could allow healthcare providers to move more quickly to a more aggressive treatment option.