By AI Trends Staff
Hospitals and doctors’ offices collect vast amounts of data on their patients, everything from blood pressure to genetic sequencing. While the data may be digitized, using it to help in patient treatment can be challenging. But the healthcare industry is getting better at using AI to find patterns in data that can help in patient care.
“I think the average patient or future patient is already being touched by AI in health care. They’re just not necessarily aware of it,” stated Chris Coburn, chief innovation officer for Partners HealthCare System, a hospital and physicians network based in Boston, in an account in WebMD. The application of AI to patient care is in an early stage and is spreading.
“I could not easily name a [health] field that doesn’t have some active work as it relates to AI,” stated Coburn, who mentioned pathology, radiology, spinal surgery, cardiac surgery, and dermatology as examples.
And of course entrepreneurs see a business opportunity. GNS Healthcare of Cambridge, Mass., offers a causal machine learning and simulation technology, that combined with its reach into “next-generation” patient data, can help determine which blood cancer patients are likely to gain the most from bone marrow transplants. The company has found a genetic signature in some multiple myeloma patients that suggested they would benefit from a transplant.
“We now have the data, the technology and the processing speed to build disease models and run computer-based in silicon simulations on every possible treatment scenario and inform the physician on the right treatment for each individual based on their biology. That’s the real power of AI,” stated GNS Healthcare co-founder Iya Kahlil. Dr. Kahlil is a physicist who co-invented the computation engine underlying products of GNS Healthcare and Via Science. Her expertise spans applications in drug discovery, drug development, and treatment algorithms that can be applied at the point of care.
Among the challenges for increased use of AI in medicine are requirements to keep patient data private, while processing huge volumes of data. While names can be removed from large data sets, people today can be identified by their genetic code, noted Mike Nohaile, senior vice president of strategy, commercialization, and innovation for Amgen, the pharmaceutical giant.
Doctors also have to guard against racial and demographic bias in the data; it’s difficult to understand and interpret the algorithms running the AI. “I don’t want to trust a black box to make decisions because I don’t know if it’s been biased,” stated Dr. Nohaile. “We think about that a lot.”
AI in Clinical Diagnoses
AI algorithms are able to diagnose diseases faster and more accurately than doctors; in particular, AI is good at detecting diseases from image-based test results, writes Terence Mills, CEO of AI.io, a data science and engineering company, writing recently in Forbes. On its website, AI.io describes its products as putting “white box AI in a nutshell” and, “If white box AI is a well-constructed house, black box AI is the foundation & framing.”
Late last year, Google’s DeepMind was able to train a neural network to accurately detect over 50 types of eye diseases by analyzing 3D rental scans, showing how effective AI technology can be at identifying real anomalies.
Early detection is key to the effective treatment of cancer, such as with preemptive measures. Certain types of cancer, such as different types of melanoma, are very difficult to detect during the early stages. AI algorithms are able to scan and analyze biopsy images and MRI scans 1,000 times faster than doctors, with an 87% accuracy rate.
Medication is ideally dispensed with precision according to the correct treatment for the patient’s diagnosis. Precision medicine depends on the interpretation of vast volumes of data.
The patient’s data, including treatment history, restrictions, hereditary traits, and lifestyle, is used to determine the most effective medication. This organization of data is a strong suit for machine learning and AI algorithms. AI data management systems are able to store and organize large volumes of data to draw meaningful conclusions and make predictions.
AI systems can browse through archives of patient data stores by hospitals and health care facilities, to assist doctors in formulating precision medication for individual patients. AI prescription systems can study the patient’s medical history and help determine the likelihood that the patient will take the medication as prescribed.