The journey for digital transformation does not occur recently. In fact, it has been going on for centuries. As Dr. Nick Patel, Chief Digital Officer of Prisma Health pointed out at the latest AIMed webinar “Raising the bar on Artificial Intelligence (AI)/Machine Learning (ML) – Pragmatic applications of digital health technologies”, peer-reviewed medical journal The Lancet advocated the use of telephone to minimize unnecessary office visits back in 1879.

In 1925, the cover of Science and Invention magazine depicted a doctor diagnosing a patient via a radio and envisioned a device which eventually enabled him to conduct video examination over a distance. As such, one should not be surprised by an eye- opening numbers of new technologies streaming and modernizing the present healthcare system, especially the opportunities driven by AI/ML within the clinical environment.

Fran Ayalasomayajula, the webinar’s moderator and Head of Population Health Worldwide, HP Inc. added when we sort of look back two years ago, in 2018, there were already more than 30% of healthcare executives who are investing in the space. By 2021, we could anticipate these investments to go beyond six billion dollars as the application of AI/ML is projected to save up to 150 billion dollars of annual medical cost in the next five years.

Thus, the focus should be on how AI/ML is being utilized to change the way we interact with the care delivery system? Specifically, the impact AI/ML has on patients, providers, and payers, the golden triangle of healthcare.


Dr. Evan Muse, Cardiovascular Genomics Lead at Scripps Research Translational Institute said the media had successfully showcased the promising nature of AI/ML in medicine but how emerging technologies disrupt physician-patient experience is less mentioned. Personally, he believes AI/ML will empower the relationship, as data continues to provide insights into areas that we haven’t look at before and powerful analytics will tell us things that our eyes can’t see.

For example, wearables and mobile health devices captured information coming from users’ diet, physical activities, sleeping patterns, and vital signs over a period of time. These details not only act as baselines for individuals to monitor their state of health and/or medical condition at home, but they could also be combined with other intrinsic factors like one’s genomics and microbiome to generate more personalized interventions shall the need arise.

Likewise, the same information can be combined with medical data obtained from electronic health records (EHRs) and traditional clinical biomarkers to generate intelligence risk-scores and provide better preventive care. Besides, large screening of EHRs, as Dr. Muse added, may better identify individuals with certain diseases and the method can also be used to predict clinically meaningful events in hospitals.

However, Dr. Muse cautioned most AI-related research were conducted on retrospective sense, there are few studies that are putting algorithms head to head in terms of what’s going on real-time in clinics. Nevertheless, he thinks AI is in a very exciting stage now and we can all look forward to the transition that the healthcare system will make. Most importantly, he indicated the importance to look at a big problem or a specific deficit, so that the right technological tools can be applied.


Dr. Patel agreed AI makes sense of data to get meaningful insights and automation will assist healthcare systems to get a clear picture of all the information. As a healthcare provider, Dr. Patel noted data navigate patients to the right avenues at the right time. At Prisma Health, they employed chatbots to engage patients and escalate their care when needed as the company believes chatbots play vital roles particularly in remote patient monitoring because it is efficient and is made available 24/7.

“I don’t have to set up a large access center or call center… It gives you real-time interactions without having a human touch-point. It’s adaptable which is extremely important in healthcare when you deal with so many different chronic disease states and there is an ever-growing content,” Dr. Patel explains.

Furthermore, it remains a norm for patients to come to the doctors’ offices for consultations. Although healthcare providers will like to spend as much time with the patients as possible, they don’t often keep in touch after the encounter. It’s challenging for doctors to find out how patients are doing beyond visiting hours. Even during a consultation, there is a computer in between the patient and doctor, there is an absent, if not minimal, seamless contact between two partiers. Dr. Patel thought it’s exactly these types of engagements that lead patients going to the emergency rooms or high acute centers in general.

Chatbots, conversely, are able to provide immediate help and guide patients when they need attention the most but Dr. Patel also emphasized this cannot be done without human interventions. That’s what he and his team have been working on, ensuring there is a human voice behind chatbots and patients are receiving appropriate assistance. For example, if a patient’s glucose is high, the chatbot will find out the reason behind and if the patient says he or she cannot afford the medication, the chatbot will alert someone from the pharmacy and enrol the patient into some scheme to get access to the medication.


Dr. Setu Vora, Chief Medical Officer of Mashantucket Pequot Tribal Nation mentioned there is a significant degree of non-value add activities like fraud, waste and abuse taking place in our healthcare system. Some studies had reported about 25-30% of all healthcare spent in the US is non-value add and even if the number is lowered to perhaps 3-5%, that adds up to a huge amount of potential savings that can be re-invested as meaningful medical or social interventions.

Again, data can play a pivotal role in reducing phantom or excessive billing, up-coding, and unbundling of services. Dr. Vora said we live in a very data-rich environment. For example, benefits claim forms or typical hospital-based claim forms contain about 27 to 38 or even more variables. At the same time, processing these claim forms is also a very complex manual process that involves over 50 steps just for examinations.

If we can leverage on AI/ML, these skill-based and cognitive demanding tasks not only would be automated but algorithms may also easily flagged any sub-optimal practice. Dr. Vora highlighted the importance of including subject matter expertise in the early phase of deploying AI/ML, he stressed that the algorithms need to be repeatedly tested and validated before it can ultimately be used on high-risk claims. After all, it’s not about re-shaping the entire system and the way we work, but also learning to re-shape our mindsets to detect every opportunity that AI/ML can be helpful.

You may revisit the webinar here.


Author Bio

Hazel Tang A science writer with data background and an interest in the current affair, culture, and arts; a no-med from an (almost) all-med family. Follow on Twitter.