Artificial Intelligence and Medicine

Advances in computer technology are reshaping the way that doctors and medical professionals approach medicine.

Today, medical professionals have access to more health data than ever before. Now, medical researchers are finding ways to use AI to quickly uncover insights in scans, vital readings and more.

This is how AI is being used right now to improve how doctors treat patients.

Using Learning Algorithms to Detect Cancer

One use of AI is finding patterns in large, unstructured data sets. CT scans and images of possibly diseased tissue are perfect examples. 

The “gold standard” method for detecting breast cancer, for example, is radiologist interpretation of lymph node samples. Radiologists look at scans of these tissue samples for signs of damaged cells, which can indicate the presence of breast cancer tumors.

The method is the best available, but it isn’t perfect. False negative rates for lymph node analysis are as low as 5 to 7 percent at best. Analyzing these tissue samples can also be a tedious process. Radiologists may spend a significant amount of time analyzing all available samples.

In 2018, Google researchers created a new algorithm, the Lymph Node Assistant (LYNA). The algorithm, which was trained on a database of healthy and diseased lymph node tissue, “looks” at tissue in the same way a radiologist does to detect possible tumors. Research found that the LYNA was even more effective than radiologists in some cases. 

Hospitals aren’t using the tech just yet, however. In the interest of patient safety and compliance with medical regulations, researchers want to make sure the algorithm is tested in a variety of settings first.

Similar AI tech has been used in the field already, however. Doctors in China applied AI to improve treatment efficiency during the height of the country’s COVID-19 outbreak. There, radiologists used AI algorithms to speed up the search for ground-glass opacities, a sign of pneumonia caused by COVID, in CT scans of patient lungs.

More Accurate, Intelligent Predictive Health Models 

Scans aren’t the only medical data that doctors are training AI algorithms with. Doctors and medical researchers are also using unstructured health data from electronic health records (EHRs) to predict patient health and responsiveness to different treatments.

In one example, researchers at the Center for Clinical Artificial Intelligence used AI to create personalized health prediction models for myelodysplastic syndromes (MDS), a group of bone marrow disorders.

Using the model, researchers were able to accurately determine patient risk for mortality from MDS. They were also able to predict whether or not the MDS would evolve into acute myeloid leukemia (AML), an aggressive form of bone marrow cancer that requires different treatment from MDS.

While the model was limited to this specific set of bone marrow disorders, the framework behind the tech is widely applicable. Other organizations, like medical startup ClosedLoop, are already working to apply the tech more broadly. The startup, rather than creating AI models for specific conditions, is looking to create a framework that doctors can use to create their own models.

The tech may help doctors make better decisions about patient care. Data analysis may show that one patient, for example, may benefit significantly from a less-used treatment than the go-to treatment a doctor typically uses. The doctor can use this analysis to inform their approach.

Artificial Intelligence Is on Track to Change Medicine

Advances in computer technology are reshaping the way that doctors and medical professionals approach medicine.

Today, medical professionals have access to more health data than ever before. Now, medical researchers are finding ways to use the pattern-finding ability of AI to quickly uncover insights in scans, vital readings and more.

Over the next few years, the tech is likely to become widely used.