12 healthcare areas where AI will result in near-term breakthroughs
Faculty members of Partners HealthCare have ranked artificial intelligence-enabled technologies that will have the greatest impact on medicine in the next 12 months.
The Boston-based health system announced the 2019 “Disruptive Dozen” AI technologies last week during a panel session at its World Medical Innovation Forum.
Also See: Partners HealthCare plans AI rollout to all its clinicians, researchers
More than 60 in-person and telephone interviews were conducted with Partners HealthCare faculty to nominate the breakthrough innovations that were ultimately whittled down to 12 AI technologies, according to Erica Shenoy, MD, associate chief of Massachusetts General Hospital’s Infection Control Unit.
To be considered for the annual ranking, Shenoy said nominated healthcare innovations had to have strong potential for significant clinical impact and patient benefit in comparison to current medical practices, as well as be on the market by 2020.
“They have to be technologies that are pretty close to making it to market,” added Shenoy. “The idea here is that these are high probability of deployment within the next couple of years.”
The 12 healthcare areas in which AI is expected to result in breakthroughs over the next year include:
1) Reimagining Medical Imaging—Researchers are using AI-based approaches to increase the power of mammography, transforming it from a one-size-fits-all method to a more targeted tool for assessing breast cancer risk. For example, a team in Massachusetts is leveraging machine learning in multiple ways to improve breast cancer screening. The researchers developed an AI-based method, now in clinical use at a large hospital for slightly more than a year, that can automatically determine breast density using mammograms.
2) Better Prediction of Suicide Risk—Research teams in Washington, Virginia and other states, as well as teams based at major social media companies, are using natural language processing and machine learning methods to create algorithms that can analyze data and detect the early warning signs of suicide. Based on their predictive power, such tools could form the basis of an app or other technology-based system that parents, other caregivers and medical providers can use to alert them when an adolescent in their care is contemplating suicide.
3) Streamlining Diagnosis—Clinical imaging has become increasingly digitized, paving the way for computer vision and other AI-based approaches that promise to improve clinical operations and decision making. One example is clinical workflows. AI offers the prospect of prioritizing patients’ images for analysis, moving them to the top of the virtual stack based on the likelihood of an abnormal and potentially life-threatening finding. For example, researchers in California recently developed a deep-learning-based algorithm that can distinguish normal chest X-ray images from abnormal ones.
4) Automated Malaria Detection—With the help of deep learning methods, a research group in Washington has developed a software program to help automate malaria diagnosis. Notably, their tool can detect and quantify malaria parasites with 90 percent accuracy and specificity, matching the level of performance of human experts. Researchers packaged their algorithm within an inexpensive, automated digital microscope, which has yielded impressive results in field tests in Thailand and Peru. The system continues to undergo further field tests and is now in commercial development.
5) A Window on the Brain: Toward Real-Time Monitoring and Analysis of Brain Health—A team of researchers in Boston has annotated some 30 terabytes of EEG data from thousands of patients. They have mined the data to create deep learning algorithms that can automatically detect seizures in the critically ill, regardless of the underlying cause of illness.
