Novel artificial intelligence (AI) platforms have made it possible for physicians to improve the clinical care they deliver, and new AI algorithms are in the pipeline that could give them a growing array of options.
Physicians and investigators at Smidt Heart Institute at Cedars-Sinai in Los Angeles, California have created an AI tool that can effectively identify and distinguish between 2 life-threatening heart conditions that are often easy to miss: hypertrophic cardiomyopathy and cardiac amyloidosis. David Ouyang, MD, a cardiologist at the institute, said these 2 conditions are challenging for even expert cardiologists to accurately identify, and patients often go for years or even decades before receiving a correct diagnosis. The new AI algorithm can pinpoint disease patterns that cannot be seen by the naked eye, and then uses these patterns to predict the right diagnosis.
Faster, More Precise Diagnosis
“AI can allow for faster and more precise diagnosis. Often, in medical imaging there are suspicious signs that pique the interest of physicians, but insufficient to definitively diagnose a disease,” Dr Ouyang said. “By using AI, physicians can be more in tune with their intuition, and bring ‘curious’ findings to the forefront to reconsider and identify challenging diagnoses early.”
In a study, the 2-step novel algorithm was used on more than 34,000 cardiac ultrasound videos from Cedars-Sinai and Stanford Healthcare’s echocardiography laboratories. By adding these clinical images, the algorithm identified specific features related to the thickness of heart walls and the size of heart chambers. This allowed them to efficiently flag certain patients as suspicious for having the potentially unrecognized cardiac diseases.
Without comprehensive testing, cardiologists find it challenging to distinguish between similar appearing diseases and changes in heart shape and size that can sometimes be thought of as a part of normal aging. The new algorithm accurately distinguished not only abnormal from normal, but also between which underlying potentially life-threatening cardiac conditions may be present. Getting an earlier diagnosis may enable patients to begin effective treatments sooner, and prevent adverse clinical events.
Dr Ouyan and colleagues reported their study findings last year in JAMA Cardiology.
Clinical trials now are underway for patients flagged by the AI algorithm for suspected cardiac amyloidosis. Many types of AI software are now being quickly developed for a host of conditions, often with the goal of enhancing patient quality of life by improving diagnosis or treatment. Dr Ouyan urged caution when adopting this technology, however.
“Too often there are AI software offers that seek to optimize billing or provide assessments that aren’t useful in clinical care,”Dr Ouyan said. “At the same time, there is often much overselling, claiming software is AI when it’s simply software and the AI is just window dressing.”
Freenome, a privately held biotech company in San Francisco, California, recently announced it is developing a tailored multi-cancer screening approach that assesses a person’s individual risk and identifies cancer signals in order to provide patients with a clear path forward. Freenome uses a multiomics platform that combines tumor and non-tumor signals with machine learning to detect cancer in its earliest stages using a standard blood draw.
“When applied thoughtfully, machine learning or AI can take volumes of data from many sources, including electronic health records, clinical studies, claims data, and process and organize the data to derive actionable insights,” said Freenome’s Chief Medical Officer Lance Baldo, MD. “For example, at Freenome, we’re developing a risk prediction tool for payers and health systems that leverages multiple layers of data to identify and surface patients who may be at a higher risk for colorectal cancer.”
This type of innovation may offer significant benefits because current risk tools for cancer screening only look at high-level demographic variables, such as age or family history. Those variables are important, but there are hundreds of other factors that could more accurately predict an individual’s cancer risk. “The rate of scientific advancement is happening so quickly that traditional methods of applying those data can’t keep pace. This is why there’s a lot of value in cloud computing, artificial intelligence and machine learning,” Dr Baldo said.
Freenome plans to enroll approximately 8000 patients in a study to evaluate their novel machine-learning approach in screening for multiple cancers.
Engineers at the University of Waterloo have developed AI technology to predict if a woman with breast cancer would benefit from chemotherapy prior to surgery. The new AI algorithm, part of the open-source Cancer-Net initiative, could help unsuitable candidates avoid the serious side effects of chemotherapy and pave the way for better surgical outcomes for those who are suitable.
The AI software was trained with images of breast cancer made with a new MRI modality called synthetic correlated diffusion imaging (CDI). With knowledge gleaned from CDI images of old breast cancer cases and information on their outcomes, the AI software can predict if pre-operative chemotherapy treatment would benefit new patients based on their CDI images.
This article originally appeared on Renal and Urology News