Artificial intelligence and machine learning technologies have been affecting our lives for some time now, from algorithms developed by Amazon, Netflix and other websites that send us intelligent recommendations on what to buy or watch next, to facial recognition on our cell phones and, as demonstrated at this year’s Google I/O, make actual phone calls to book our appointments.
AI is now poised to change the very face of healthcare. Over the last few years, a burgeoning sector of companies developing AI tools in healthcare has emerged, bringing the potential to diagnose illnesses faster and even save lives. AI tools, consisting of hardware and software capable of processing large amounts of data and delivering analysis in minutes or even seconds, can be trained to analyze almost anything, suggesting unknown possibilities in the quest to cure cancer, develop drugs for rare, deadly conditions or stem the tide of chronic disease. These systems can sift through the large volume of data that medical devices collect, in ways that haven’t been possible before. With more thorough data analysis, new patterns can be discovered which could identify a dangerous anomaly or a promising therapy. As well, AI software learns over time. The systems become smarter and thereby deliver more accurate results and new insights to a physician or researcher.
IDC predicted that worldwide spending on artificial intelligence and cognitive computing technologies will reach $19.1 billion in 2018 and leap to $52.2 billion by 2021. Studies are starting to show real-world impact from these advanced analytics systems. Researchers from Google used its deep learning algorithms to detect a chronic condition, diabetic retinopathy, with greater than 90% accuracy. Another Google project involved a system to detect cancer spreading to adjacent lymph nodes, which showed an accuracy of 89 percent, compared to 73 percent using other methods.
Preventive medicine is another promising area for AI technologies. Annual screening tests can be downright uncomfortable and inconvenient for patients. AI technologies are presenting new solutions for these important exams. For instance, mammograms are an unpleasant, angst-ridden exercise. If a patient has dense breast tissue, this can be even more stressful if the mammogram result later indicates a vague area of concern to the radiologist. The patient then must come in for another visit to rule out a possible tumor. Delphinus Medical Technologies is developing Softvue™, a flatbed ultrasound machine that incorporates AI software. Not only is this more comfortable for patients (no compression necessary) but it is also able to spot cancer cells earlier and faster or rule out cancer without the need for repeat testing.
Researchers are also looking at how AI can help in mental health diagnosis and treatment. In a proof-of-concept study reported in the journal Nature, developed an AI model with the goal of accurately predicting the development of psychosis by analyzing the speech patterns of at-risk youth. Though it was a small study, the early results are promising. From early diagnosis to virtual therapists, AI is starting to make a big impact on mental health science.
In the near term, artificial intelligence could help the most by enabling doctors to be more efficient. Primary care physicians are hampered by cost constraints with the mandate to improve outcomes in the age of healthcare reform. Overworked doctors and P.A.’s must handle more patients as health insurance companies adopt the gatekeeper model. Technology companies are developing tools which crunch the medical records of patients, reviewing historical data and incorporating the latest research to help arrive at better decisions, with far less manual work. With better accuracy and testing through AI, Accenture estimates that AI could potentially save the U.S. economy $150 billion annually in health care costs by 2026. So, the adoption of AI is not only mandatory to keep up with a congested healthcare system, it’s also groundbreaking.
At its core, Artificial Intelligence systems depend upon high-performing hardware to support the rapid analysis of large data sets. Advances in hardware technology are finally making it economically viable to support these life-changing platforms. NVIDIA is at the forefront of the AI industry, starting with the Nvidia DGX-1 system which delivers up to three times faster training speed than other GPU-based systems. This speed, combined with CPU capacity of 25 racks of conventional servers integrated with an advanced deep learning software stack, means that healthcare organizations can simplify and fast-track AI projects.
The latest technology, Nvidia DGX-2, brings even greater performance and scale. Engineers can train models 4X bigger on a single node. In comparison with legacy x86 architectures, 1x DGX-2’s ability to train the deep neural network ResNet-50 would require the equivalent of 300x servers with dual Intel Xeon Gold CPUs and a price tag of over $2.7 million dollars. The Nvidia solution is a fraction of the cost, space and power requirements of any other solution on the market today.
As one of Nvidia’s Elite Partners, we at GPL Technologies bear witness to the AI paradigm shift in healthcare, architecture, visual effects (VFX) and other sectors on a daily basis. We know, with its latest DGX-2 and GV100 cards, Nvidia has both mindshare and market share in AI technology and expertise that pushes them years ahead of the competition. The future of healthcare depends upon intelligent technologies which when blended together at the point of care, will reduce the waste, error, cost, and inconvenience that have plagued patients and providers everywhere.
If you’d like to learn more about AI solutions or if you have any questions regarding the tech, write to us at email@example.com.