Image credit: Lambert Leong
The pandemic put pressure on healthcare facilities that is only now beginning to ease. Hospitals and medical staff patients were forced to find innovative ways to care for patients, and that innovation continues as medical facilities evaluate which of those efforts will help their financial recovery and boost patient care.
Integrating artificial intelligence into patient care is one appealing approach. Hospitals have large data sets, which AI algorithms can evaluate to identify medical insights and offer predictive analysis. In fact, the University of North Carolina Lineberger Comprehensive Cancer Center used IBM’s Watson for Genomics to suggest cancer treatments for patients, well before the pandemic.
AI Suggests Cancer Treatments
In the initial study, Watson analyzed genetic information, past studies, and databases to identify treatment options for more than 1,000 patients. IBM’s Watson for Genomics solution was able to suggest treatments for 700 patients with genetic abnormalities, which matched the findings of a panel of medical professionals.
Watson also identified “potential therapeutic options” for an additional 300 patients that the panel of doctors missed. Of those, nearly 100 were singled out for treatment based on only eight genes, which the panel of doctors didn’t consider actionable.
Watson later came under fire for suggesting unsafe treatment plans. Researchers maintain no patients were harmed, and the Watson recommendations were just one of several tools medical staff considered when formulating patient treatment plans.
Even if Watson’s initial treatment options weren’t ideal, the research proved that AI can be an essential tool in driving patient outcomes. The sheer volume of medical research makes it impossible for doctors to keep up with every new study and finding. AI algorithms can find and analyze this information and present summaries to clinicians, keeping them updated quickly and frequently.
AI-driven insights will improve medical care because they provide clinical decision support (CDS). Deep learning algorithms are able to cull through massive amounts of patient data. These solutions can identify patterns and provide insights that medical staff and researchers simply can’t process as effectively. This kind of insight will drive precision medical care, enabling doctors to customize treatment plans based on patient-specific genetic information and overall trends.
Diagnosing Cancer through AI
Researchers at Tulane University developed AI algorithms that can identify and diagnose colorectal cancer based on tissue scans. Pathologists can tap into this technology to meet increasing demand without fear of misdiagnosis. The AI scored slightly better than pathologists when identifying colorectal cancer. Researchers hope that enabling quicker diagnoses will lead to earlier treatment.
Genome-based treatments enable doctors to look at tumors in a new light. Typically, cancer is classified by its location: lung cancer, breast cancer, colon cancer. With DNA sequencing and artificial intelligence, tumors can be analyzed and assessed based on their characteristics. Armed with that information, clinicians can develop personalized treatment plans and try drug therapies that might fall outside the standard approach to treatment.
Researchers at the University of Hawaii have discovered that deep learning can indicate which women will or won’t develop breast cancer based on their mammograms. The research team, using more than 25,000 digital screening mammograms, successfully trained a deep learning model to detect details in the mammogram images of women who developed screening-detected breast cancer.
The team suggests that radiologists can use AI as an alternative to additional imaging, which is expensive and can delay treatment. By classifying women as having low or high risk factors, subsequent mammograms can be scheduled accordingly.
Similar technology from MIT is being used at Massachusetts General Hospital to identify women with a high risk of developing cancer. The technology was launched during the pandemic, when patients were skittish about visiting medical facilities for breast cancer screenings. Several women were diagnosed with cancer after they were persuaded to get mammograms, based on prior images.
- See how Intel’s work in AI is helping healthcare professionals advance precision medicine.
- Learn more about how IBM Watson is used today.