AI and disease treatment

From immunotherapy to robotic surgery to stem cell transplants, cancer treatments have advanced by leaps and bounds in recent decades.

Still, every patient is different and reacts to treatments in different ways. What works for one patient may not work for another. While one patient may recover quickly from surgery or a transplant, another may develop complications.

New AI tools may make it more possible to better predict treatment outcomes.

It has the ability to analyze large data sets, find correlations difficult to identify by current methods and analyze genetic mutations. This allows AI to help doctors make sure treatment protocols are tailored to specific patients while minimizing risk of side effects and avoiding treatments that may be ineffective.

For example, immune checkpoint inhibitors have produced good outcomes for millions of patients with difficult cancers, such as melanoma. But these drugs are less than effective in more than 50 percent of patients.

Researchers at the National Cancer Institute are developing an AI tool designed to help predict which patients may respond to checkpoint inhibitors and which ones may not. Among the factors the algorithm analyzes are a patient’s age, cancer type, treatment history, blood albumin level and blood neutrophil-to-lymphocyte ratio. This tool may eventually help increase the number of patients who will benefit from checkpoint inhibitors.

Doctors at City of Hope have developed an AI model designed to predict which stem cell transplant patients may develop sepsis, a dangerous and fast-growing blood infection. Up to 10 percent of transplant patients develop sepsis, which comes on rapidly with few early symptoms. Patients who develop sepsis often have poorer treatment outcomes than those who do not get the infection.

City of Hope doctors also have developed an AI model to help predict which surgical patients may develop complications, such as bleeding, respiratory or cardiac distress or poor wound healing.

“Having a better ability to predict who is at risk for postoperative complications can help with preoperative optimization of the patient, as well as improve the consenting process for the patient and family,” says Lily Lau Lai, MD, a colorectal surgeon and professor at City of Hope.