From faster diagnosis to enhancing operations across care management, compliance and accounting, artificial intelligence (AI) is revolutionizing the healthcare landscape. While AI is not a new concept, its adoption in the healthcare industry has lagged behind because of significant changes in technology and the exponential growth of data. AI has tremendous application for healthcare, but leaders will need to balance data management and governance to properly enhance the patience experience and outcomes through AI implementation.
The following is a conversation with Deana Rhoades, health plan solutions lead, NTT DATA Services, on the topic of robotics and AI in healthcare.
How is NTT DATA Services helping its healthcare clients improve their customer care and services with artificial intelligence? Can you provide some specific examples? At NTT DATA Services (NTTDS), we are continually seeking ways to utilize AI to facilitate the delivery of care to patients and/or members. Our perspective at NTTDS is that data and intelligence are at the center of healthcare. This data is consumed by various solutions in AI space, including machine learning, natural language processing, neural networks, etc. At this stage, the data has been transformed into actionable insights that can be acted on by others, such as RPA (bots), virtual agents or even humans. An example of this is the creation of a predictive model using machine learning (a subcategory of AI) that identifies members likely to become high risk in the next 12 months. The list of members, along with the appropriate supplemental information, can be utilized by a bot or virtual agent to engage with the member in an effort to mitigate their rising risk. The supplemental information is used to provide the appropriate action for the member to take to better manage their health and reduce their risk. All of this can now be accomplished near-real time rather than retrospectively through use of AI.
What are some of the most significant ways artificial intelligence can benefit the healthcare industry? How is it being applied to healthcare? How is it currently falling short?
AI-based solutions have impact across the healthcare continuum — from the patient/member to the provider to the health plan. Some examples include:
Managing prescription drug costs: Use of predictive models to identify use of high cost drugs when generic equivalents are available. AI uses the output of these models to make recommendations to providers in their Computerized Physician Order Entry (CPOE) systems as to which drugs would be most appropriate to prescribe to a particular patient. CPOE first came into use when providers started sending electronic prescriptions to pharmacies. Now it’s widely used to order other services such as radiology, labs, etc. Through this effort, health plans are able to work with providers to proactively manage claim costs and help to lower out of pocket expenses for members.
Proactive identification of care gaps for chronic members: Using machine learning and natural language processing, predictive models have been created that identify members who have a higher likelihood of having care gaps by incorporation of social determinant of health indicators in the model. As with the scenario described in the answer to the first question, this model has resulted in better, more prescriptive outreach to members that addresses the underlying reason for the care gap (e.g. lack of transportation to appointments).
IoT for medical devices: Now regulated by the FDA, there are two devices that have been approved for use in 2018. One is an AI-based software program that is able to provide insights relative to retinopathy and make decisions/recommendations based on those results, without intervention from an clinician. This solution is called IDx-DR. The second is a device analyzes images of blood vessels in the brain and send alerts to first-line care professionals who can then act quickly to offset complications for potential stroke patients.
“Smart” virtual assistants: Healthcare is turning more and more to virtual assistants because of a shortage of nurses and other caregivers. NTT DATA’s Pepper is one such example. This robot uses both AI and robotics to provide companionship and feedback to members/patients. Based on data it has about a member, Pepper is able to provide medication or appointment reminders. In addition, Pepper collects information regarding the member/patient well-being and provides that data to others for action.
Improved imaging analytics: NTTDS has used machine learning and deep learning to create algorithms that return results of images much faster than a human can.
Putting the power of healthcare in the patient’s hands: An example of this is a mobile phone app called CaptureProof. A form of telehealth, this app uses AI to capture images from the patient that are sent to the provider for review along with results generated from embedded algorithms that assist in determining the patient’s stage of recovery. The app also provides reminders to the patient regarding what actions to take next, such as the appropriate exercise for their stage of recovery from knee surgery (www.captureproof.com). This is just one example of patient/member empowerment; many more exist in the industry.
I don’t know as we can say that AI is falling short for health care, as there are new applications of AI in healthcare that are being developed every day. There are possibilities for the application of AI in health care that haven’t even started to be explored, and probably many more that haven’t been thought of yet. What we do know is that as the amount of healthcare data available continues to grow at exponential rates, there is an immense opportunity to improve the outcomes for members/patients, reduce the potential for errors, and reduce the overall cost of healthcare.
What do you think are some of the biggest obstacles preventing hospitals and other health facilities from implementing artificial intelligence? AI is not a new concept; it’s been around for decades. The application of AI in healthcare has been slower on the uptake as the healthcare has struggled with how best to handle the vast amounts of and increased accessibility to healthcare data that has occurred with the implementation of EMR/EHRs, laboratory information management systems, practice management systems, PACS, etc. The amount of healthcare data is going to continue to grow at an exponential rate as more and more sources of data are identified as being relevant to and/or influencing a patient/member’s health and care (think social determinants of health). As part of implementing AI, those in the healthcare industry need to have robust data management and governance strategies in place. This will ensure that the outcomes from the application of AI are of the highest quality and accuracy possible. The old adage “garbage in, garbage out” still applies today. Another reason that healthcare organizations may be hesitant to implement AI is that AI is a disruptor in healthcare; it has the capability of changing the healthcare paradigm. As a result, there is careful consideration to the impact that implementation of AI will have on the delivery of care and to the patient/member themselves.
Do they understand its capabilities? Can the capabilities of AI ever be fully understood given that there are constant advances in the technology and associated applications? Healthcare organizations are very cognizant of the potential of AI and, as noted before, carefully assess how AI could impact the patient/member experience.
Why do you think emerging technologies, such as artificial intelligence, are so important for the healthcare landscape? What are some of the other important emerging technologies? With the continued shortage of healthcare providers, AI is likely to play a critical role in closing the gaps. In addition, with IoT devices now being able to supply near-real time information regarding a patient/member’s condition, it becomes possible to predict and/or prevent catastrophic events, or at least mitigate the severity of such an event. Through mechanisms such as these, the quality of outcomes will increase while lowering the overall cost of care. For example, through use of the device that analyzes blood vessels in stroke patients: The early identification and treatment of potential complications is going to improve the overall outcome for the patient as well as lower the overall cost, as the patient won’t incur the costs of a longer hospital stay and/or more extensive treatments.
What’s getting too much hype right now that you think won’t last? This is an interesting question that is difficult to answer. As noted previously, there are constant advances in healthcare from both a process and a technology perspective. Identifying which of these will be integrated successfully by the healthcare industry depends on several factors: a) the impact on the industry as a whole (e.g., does the solution drastically reduce the cost of care), b) the impact on the patient/member (e.g., improved outcome or increased patient/member engagement), and c) the ease in which a new solution can be adopted. It’s a difficult balancing act for healthcare organizations, which makes it difficult to predict which advances will be successful and which will not.
How do you predict this technology will continue to integrate itself within the healthcare field in real terms? Keep in mind that AI is a very broad term that covers multiple different technologies as described in earlier answers. At NTT DATA, we see patient/member empowerment as a driver for adoption of new and emerging solutions.
What are some of the tangible returns from boots on the ground? Refer to bullets 1 & 2 above, as they speak specifically to what we’ve been able to do for our healthcare clients.