Predictions Series 2022: AiThority Interview with Dr. Arnaud Rosier, CEO & Founder at Implicity

Hi, please tell us about your journey in AI technology and how you started with Implicity.

I founded IMPLICITY about six years ago to solve some of the workflow challenges I faced as a Cardiac Electrophysiologist. My Ph.D. is in health informatics, and I wanted to develop a tool that would enable the adoption of remote cardiac patient monitoring. So, we developed a cloud-based platform leveraging machine learning and AI-based algorithms to aggregate data from various cardiac implantable electronic devices (CIEDs) and manage the entire clinical and administrative workflow. This allows cardiologists, electrophysiologists, and their teams to scale remote patient care and optimize revenues.

Today, Implicity covers more than 70,000 patients in 100 medical facilities across the United States and Europe. In addition, the platform is used by clinical and academic researchers. Investigators can use intelligent cardiac remote care and research solution to support their studies.

Recommended: AiThority Interview with Dario Debarbieri, Head of Marketing at HCL Software

What kind of problems does Implicity solve for users? What data science and AI skills should users have to leverage your platform?

Healthcare providers – and cardiac teams, in particular – have increasingly turned to remote patient monitoring (RPM) in recent years to help detect potential health problems in patients before they become more serious. A cardiologist might track metrics such as physiologic monitoring (i.e., weight, blood pressure) from external devices and feeds from devices such as pacemakers or implanted loop recorders. But remote monitoring also challenges clinicians despite the opportunity to deliver care proactively.

Previously, most remote monitoring devices pushed out data to proprietary hardware. But increasingly, smartphone apps are collecting patient data and sending it out to the public cloud. So, hospitals and clinics have this wealth of data, but often it’s sitting in several different places outside the walls of their organizations, being stored and managed by third parties. To successfully run analytics on this data, healthcare providers must bring it into one unified environment through a data management platform. This is where Implicity comes into play. By streamlining data access, our AI-powered platform makes remote monitoring a more practical option for electrophysiologists because they can analyze the flood of data and information from remote devices without data science or AI expertise.  The solution enables clinicians to monitor patient data continuously and intelligently prioritize health events that require further action.

One of the major hurdles preventing the widespread adoption of remote monitoring is the volume of false positive alerts. Up to 90 percent of signals from RPM systems are mere “noise” that does not require intervention. Monitoring devices are designed to be extremely sensitive, so that no critical event is missed. But it can be extremely taxing for clinicians to monitor these systems and weed out false positives constantly. Implicity is employing the power of artificial intelligence to help analyze ECG data from ILR technology in hopes of assisting electrophysiologists and their teams to avoid wasting time reviewing these false positives. Our AI-powered platform can do this work on behalf of clinicians, allowing them to focus their attention on the patients who genuinely require care.

Why suddenly do we see rapid adoption of AI and machine learning in the healthcare industry?

The demand for more personalized care is one of the most significant factors driving the adoption of AI in healthcare.  For decades, people have fantasized about a future in which AI tools could synthesize a patient’s clinical data to create tailored wellness plans and help predict adverse medical events. The problem is that much of this data (for instance, doctors’ notes) is unstructured, making it extremely challenging to process all relevant information with AI.

However, thanks to smartwatches and fitness trackers, there is now more structured health and wellness data about patients than ever before. It’s important to note that these devices don’t merely collect data on things like physical activity and sleep but also on, more specifically, actionable clinical data. For instance, Apple received FDA approval for its built-in ECG feature on the Apple Watch.

Right now, most health tracker data never reach healthcare providers (instead of remaining on devices or getting pushed to vendors’ cloud-based platforms). Soon, however, patients will expect their healthcare providers to access their health data and deliver personalized care plans. This is one of those things that will likely happen slowly at first and then all at once. Similarly, for twenty years, facial recognition and voice recognition applications failed to work as technologists worked to fix their problems. Then, practically overnight, they became integral parts of daily life.

Recommended: AiThority Interview with Wayne Busch, Group President at NTT DATA Services

Which domains within the healthcare industry is AI widely used and why? Which parts are still lagging?

Generally, the most adapted industries to AI are those for which digitized, complex data modalities are used for humans. As we can see with applications like Google on facial recognition, AI lends itself well to recognizing shapes in images. Medical imaging is, in essence, an area where AI is developing a lot because it takes a long time to train a radiologist. And in most cases, AI can do just as well. AI is also suited for fields such as histology because scanned images and samples are taken to detect cancers, etc.

But AI will eventually be beneficial in the field of connected medical devices. Physicians are waking up to the need to offer remote monitoring because it decreases mortality, but they have difficulty scaling their programs. So, we’re going to see more use of remote monitoring for implantable devices like pacemakers and wearable devices like weight scales and blood pressure monitors. But patients need to know that physicians are reviewing the data for them to adopt these wearable devices, and that’s impossible without AI.

We’re also hoping to see artificial intelligence reduce burnout among clinicians. Job burnout is a huge issue in healthcare, especially for cardiologists. They’re working long hours, putting themselves under enormous pressure to be perfect in their work because it can mean the difference between life and death for their patients. A 2020 report from McKinsey talks about how AI can improve practitioners’ lives because it lets them spend more time looking after their patients. And then, as a result, you see improved retention and staff morale. So, as the industry adopts AI more widely, that’s one of the significant benefits we’re hoping will play out.

Recommended: AiThority Interview with Paul Sobel, President and CEO at Wodwo

Your take on the influence of Ethics and Biases in AI applications — how do you neutralize these talks?

COVID-19 has put a spotlight on the need for health equity. AI may worsen disparities if nothing is done to prevent bias in algorithms. Algorithms are susceptible to ethnic, gender, and social discrimination, which has significant implications.

AI can worsen the quality of care when algorithms are trained with data from only a specific typology of the population.  Preventing bias in health care requires large quantities of quality data. An example may be the National Health Insurance database in France. France is a country based on a universal healthcare system, and the dataset is an unbiased representation of the whole French population, regardless of social status, age or ethnicity.  Health Data Hub is a health data platform put in place at the end of 2019 by the French government to combine existing health databases and facilitate their usage for research and development purposes.

Implicity is the laureate project of a health data hub with “Hydro,” a complete deep-tech solution to predict heart failure and prevent aggravation and hospitalization. Using a universal dataset of patient records and not only the patient records of some specific hospitals make our AI-based solution unbiased and less likely to cause inequalities in healthcare.

Thank you, Arnaud! That was fun and we hope to see you back on AiThority.com soon.