Intel has developed a suite of AI technologies and has been collaborating with numerous medtech providers to create new healthcare solutions based on data-driven strategies. AI has come on in leaps and bounds, and is beginning to make an impact in various healthcare fields. Intel aims to be at the forefront of this AI revolution.
For instance, Intel has collaborated with Novartis to perform high content drug screening. The company uses Intel neural network technology to analyze thousands of images of cells to identify promising drug candidates. Previously, technicians analyzed these images manually, which was tedious and time consuming, so the new technology is very helpful. It is also much faster, reducing the screening process from 11 hours to just 31 minutes.
AI systems are proving to be very adept at image analysis. In another application, Intel worked with an AI company called MaxQ AI to create AI technology that can analyze CT images of stroke and head trauma patients. At present, there is an enormous error rate in CT image analysis for intracranial hemorrhage, and about 30% of all diagnoses are incorrect. So far, MaxQ has tested the technology in a preclinical trial, and demonstrated an impressive 97% sensitivity and 90% specificity.
In another application, Intel helped AccuHealth, a chronic care management company based in Chile, to develop an AI monitoring system. Data from biometric sensors that patients wear at home are collected and analyzed by the AI software. The system can anticipate health problems that could lead to an emergency room visit. Since the company has begun using the system in Chile, it has reported that emergency room visits have decreased by 42% in monitored patients while reducing costs to insurers by up to 50% per patient.
See a video about Intel AI in a hospital context which describes another example of how the company helped Montefiore Medical Center in the Bronx predict who was at risk for respiratory failure, and prevent it from happening, below:
Medgadget had the opportunity to ask Chris Gough, GM of Health and Life Sciences Group at Intel some questions about Intel’s AI technology.
Conn Hastings, Medgadget: Intel is best known for its computer chips. How and when did the company get involved in AI and healthcare?
Chris Gough, Intel: Intel’s work in healthcare is not new and has been an important focus for decades. Our technologies are widely used throughout the industry to help unlock new possibilities with a comprehensive stack of products designed for AI. These products get deployed across the data center and the cloud through to the network and within embedded systems at the edge. We feel that AI will find patterns in data that the human eye just can’t see, finally enabling the healthcare industry to drive insights from the massive amount of data that largely goes unused today. It is transformational and is already helping the industry address their biggest challenges, and Intel is at the forefront of technology innovation that is making this happen.
Medgadget: Please give us an overview of the various healthcare applications in which Intel AI has been applied.
Chris Gough: AI is already delivering measurable results for clinical, operational and financial use cases. We’ve also seen a lot of focus on using AI to drastically improve drug discovery, medical imaging and risk stratification as three examples.
Medgadget: How has AI helped in these contexts?
Chris Gough: On the clinical side, GE Healthcare’s 510(k) FDA pending Critical Care Suite will help radiologists cut through the clutter to find the truly urgent cases that if not read immediately could negatively impact patient care. Today x-ray exams are 60% of all images performed and the majority are routed as STAT even though we know it’s impossible for radiologists to prioritize everything as urgent. ‘Pneumothorax’ is one of those high priority cases when air leaks into the space between the lung and chest wall pushing on the outside of the lung and making it collapse. Patients who present with symptoms associated with this condition receive a chest X-Ray, which can take anywhere between two to eight hours to read, and that delay can be fatal for somebody battling this. New algorithms will identify this potentially life-threatening condition with high accuracy within seconds at the point of care.
On the operational side, a great example is our work with Sharp HealthCare, who established specialized rapid response teams (RRT) that respond to medical emergencies in the hospital. As time permits, these teams may manually review charts to spot a potential crisis. Sharp wanted to see if predictive analytics techniques, including artificial intelligence methods such as machine learning, could automate that analysis and help the hospital deliver more proactive, efficient interventions. When Sharp tested the model against historical data, the model was 80 percent accurate in predicting the likelihood of an RRT event within an hour, demonstrating the potential to drive real-time clinical interventions, improve outcomes, and enhance resource utilization which is an operational efficiency goal of theirs.
Finally on the financial side, we helped Chilean-based chronic care management company AccuHealth use predictive AI to identify warning signals captured from biometric sensors worn by chronic care patients. This way, AccuHealth could catch health problems before patients have to rush to emergency rooms. According to AccuHealth, the company’s services in Chile have decreased emergency room visits by 42% while saving six of the largest insurers an average of 50% per patient.
Medgadget: What has been your experience in using AI to predict serious illnesses, such as acute respiratory distress syndrome or sepsis? Has the Intel AI system met with success in this sphere?
Chris Gough: Absolutely. Montefiore’s PALM platform (Patient-centered Analytics Learning Machine) which is powered by Intel technology is a great example and is proving itself out in the Intensive Care Unit helping doctors save lives. PALM flags patients headed toward respiratory failure. We know that there are medical breakthroughs that remain elusive because those insights are trapped within data silos. These disparate databases with different nomenclatures or other structural incompatibilities prevent them from being pulled together. PALM doesn’t care if the data are in traditional databases or in newer unstructured data stores that might contain voice, images and sensor inputs. PALM is able to do this through a semantic data lake architecture that combines Montefiore’s multiple stores of medical data with various ontological databases that define over 2.5 million terms and their relationships to one another.
Additionally, PALM integrates an ‘Analytic Tapestry’ that detects and connects algorithms doing similar work (for example, reducing hospital readmissions) in order to improve algorithm performance. Finally, PALM is ready for what may come next – it can quickly and easily integrate new data and ontologies, including information like voice, images, and sensor inputs from internet-of-things (IoT) devices.
PALM will help Montefiore predict and prevent a variety of costly life-threatening conditions, ranging from sepsis to heart failure, to improving operations by predicting appointment no-shows, or to guiding walk-in traffic to urgent care clinics and emergency rooms with better availability.
Medgadget: Do you think that much of the data that is collected about patients are wasted unless analyzed using an AI system?
Chris Gough: Yes. 80% of all healthcare data are unstructured data today (i.e. clinical notes, socio-economic data, X-Rays, etc.) and an even larger percentage of healthcare data goes unused. It’s just too hard to wrangle it and make sense of it without AI. AI is the prescription the healthcare industry needs to start finding life-saving patterns in these data.
Medgadget: What are your future plans for this technology? Do you envisage that AI will drastically change the way patients are cared for?
Chris Gough: We envision a future era of ‘precision health’ where the healthcare industry can become significantly more proactive vs. reactive. In the future, the healthcare organizations that use AI and leverage their data sets to predict, preempt and prevent problems before they become acute will have a significant advantage over their technology lagging competitors. AI will enable leaders in the industry to provide the right intervention to the right patient at the right time. We’re excited to be accelerating this transition to lower costs, improved access, and better outcomes in the health and life sciences sector.
Link: Intel AI in Healthcare…