The Miller School of Medicine’s radiology department is working with the university’s Institute for Data Science and Computing to design an artificial intelligence tool that could help them diagnose patients in a more individualized way.
Over the years, new technologies have helped radiologists diagnose diseases on a multitude of medical images, but they have also changed their work.
Whereas in the past these doctors spent more time talking with patients, today they spend most of their time in the reading room – a darkened space where they review images alongside a patient’s electronic medical records. patient and other data sources – to diagnose a disease.
The profession of radiologist is often solitary. And it’s a trend that University of Miami Miller Medical School radiologists Dr. Alex McKinney and Dr. Fernando Collado-Mesa hope to change.
The two doctors worked with the University’s Institute for Data Science and Computing (IDSC) to create an artificial intelligence toolkit that will draw on a massive database of anonymized data and medical images. to help doctors diagnose and treat diseases based on more than imaging data. but also taking into account a patient’s unique background and circumstances. This would include risk factors, such as race and ethnicity, socioeconomic and educational status, and exposure. Doctors say it’s a necessary innovation in an era where narrow artificial intelligence in radiology is only able to make a binary decision such as positive or negative for a disease, rather than looking for a myriad troubles.
“We believe the next iteration of artificial intelligence should be contextual in nature, which will take into account all of a patient’s risk factors, lab data, past medical data, and help us track the patient.” , said McKinney, who is also the president. from the radiology department. “It will become a form of augmented interpretation to help us care for the patient.”
According to Collado-Mesa, “this toolkit will not just say yes or no, disease or no disease. He will point to the data around him to examine a variety of issues for each individual patient, to put his findings into context, including future risks.
Current artificial intelligence tools are also limited to a specific type of medical image and cannot, for example, analyze both MRI (magnetic resonance imaging) and ultrasound. Additionally, the patient data used in these diagnostic tools typically does not include a range of demographic groups, which can lead to bias in care. Having a tool that draws on the examples of millions of South Florida patients, while preserving their privacy, will help radiologists be more efficient and comprehensive, McKinney noted.
“Right now there is so much data that radiologists have to sift through. So that could help us as a technology partner,” McKinney added.
All of these factors led Collado-Mesa and McKinney to try to create a better alternative, and they spoke with IDSC director Nick Tsinoremas, also a professor of biochemistry and molecular biology. Tsinoremas and the advanced computing team at IDSC came up with the idea of using an existing tool, called the URIDEa web platform that aggregates anonymized patient information for academic research, and adding the anonymized images from the Department of Radiology.
They hope to unveil an early version of the toolkit this summer and plan to add new items as new imagery data is added. It will include millions of CT scans, mammograms and ultrasound and MRI images, as well as X-rays, McKinney pointed out.
“We don’t want to rush this because we want this to be a robust, high-quality toolkit,” said Collado-Mesa, associate professor of radiology and breast imaging, as well as head of innovation and artificial intelligence for the department. of Radiology.
The doctors and Tsinoremas hope the artificial intelligence tool will help answer vital research questions, such as: what risk factors lead to certain brain tumours? Or what are the most effective treatments for breast cancer in certain demographic groups? It will also use machine learning, a technique that constantly trains computer programs to use a growing database, so it can “learn” the best ways to diagnose certain conditions.
“Creating this resource can aid in diagnosis and will enable predictive modeling of certain diseases, so that if a person has certain image characteristics and clinical information similar to other patients in this database, physicians could predict the progression of a disease, the effectiveness of their drugs, etc.,” Tsinoremas said.
To ensure the toolkit will be unbiased, the team also plans to add more images and data from all population groups in the community as they become available, as well as constantly and systematically monitor the different elements in the toolbox to ensure that it is working properly.
Radiologists plan to focus first on diseases that have high mortality or prevalence in the local population, such as breast cancer, lung cancer and prostate cancer, and add others with time.
The technology could allow them to spend more time with patients and offer more personalized and accurate care based on the patient’s genetics, age and risk factors, according to the two doctors.
“Artificial intelligence has the potential to be patient advocates, rather than a one-size-fits-all approach to medicine based on screening guidelines,” McKinney said. “It could help us out, and it would hopefully offer more hope to people with rare diseases.”
But as more data is added in the future, the researchers hope to expand their work with the tool. And they hope doctors at the University will also use it to conduct medical research.
“This is a resource that any UM researcher could potentially access, provided they have the approvals, and it could trigger a number of different research investigations to describe the progression of the disease and how the patients respond to different treatments in a given period of time – these are just some of the questions we can ask,” Tsinoremas said.