“Artificial intelligence is just a tool, and one we’ve been using for a long time,” says Nassib G. Chamoun, founder and CEO of the Boston-based Health Data Analytics Institute (HDAI). “We are far, far away from AI making independent health care decisions. Today, anything that doesn’t have a human in the loop or in charge of the final decision is on shaky ground.”

Chamoun, inventor of the Bispectral (BIS) Index™, the first clinically validated direct measure of the effects of anesthetics on the brain, has been using AI in health care applications since the mid-1980s. At that time, a chance interaction with an anesthesiologist at the Harvard School of Public Health changed the course of his career.

“My vision was to use this information to build an AI platform that can summarize and synthesize information to help clinicians with the burden of electronic health records (EHRs).”

“I was pursuing a PhD in biomedical engineering at Boston University, after achieving a bachelor’s in electrical engineering and a master’s in computer engineering,” says Chamoun. “It was exciting because we were one of the first labs to use artificial intelligence in health care. Digital Equipment Corporation (DEC) shipped five specialized computer systems to labs around the country, and I was using one of these computers to adapt open-source military software to process biological signals at the Lown Cardiovascular Lab at Harvard.”

That changed when fellow researcher, the late Alex Mills, MD, an anesthesiologist, invited Chamoun into his OR at Massachusetts General where he had a patient connected to the electroencephalograph (EEG). Chamoun spent several days observing anesthesiologists at work and became fascinated with the concept of brain monitoring. He realized interpreting heart signals was relatively easy, but brain signals are much more complex. The problem was fascinating: How do you track levels of consciousness and sedation by analyzing EEGs?

“I was convinced I could come up with a solution. I was so young and naïve I thought it would only take two years and maybe a couple million dollars,” he says.

With the blessing of his mentor, the late Bernard Lown, MD, a Nobel Peace Prize winner, he dropped out of graduate school to found Aspect Medical Systems. Not only did that mark the beginning of his entrepreneurial career, but it started him on a lifelong pursuit of using technology to help clinicians improve patient outcomes. In the case of the BIS Index, he embarked on a 25-year, $250 million odyssey to leverage AI technology to extract signals from the EEG and turn them into a measure to help anesthesiologists track changes in consciousness during a surgical procedure.

Developing the BIS™ technology was an incredible journey, he says. “It took hundreds of studies and many, many researchers to develop a usable algorithm. In fact, it was 11 years before we had a viable product.” The process was guided by published research stating that when EEGs were reviewed by clinicians, only 50% agreed on the interpretation. Anesthesiology departments are very hectic workplaces, and the goal of Chamoun’s team was to develop a technology that would automate the interpretation of EEGs and reduce it to a number that clinicians could easily understand.

“We had two big things going for us,” he recalls “First, at the time, investors were interested in anything involving medical devices. Second, anesthesiology as a specialty is very innovative. From the very beginning, we had anesthesiologists who were willing to help us conduct research on the algorithm.”

To accomplish this task, they leveraged signal analysis techniques, which are effectively AI, to reduce the brain waveform to multiple pieces of information that they combined into a single number. He noted that, although BIS is a single number, it represents a combination of parameters that summarize the brainwave under anesthesia. Clinical work aimed to validate how to best combine multiple variables into an index that is effective, usable, and consistent.

“Steve Shafer and Donald Stanski, then both professors of anesthesiology at Stanford University, really helped us understand the pharmacology of anesthesia and how it altered brainwaves. From this, we could develop an algorithm that would consider how different anesthetics affect patients.”

Achieving a viable product was only the first milestone. Next, they had to get U.S. Food and Drug Administration (FDA) approval. “This type of device didn’t exist in the marketplace, so it took years of going back and forth with the FDA to get approval to commercialize the product. We designed our first trial based on whether a patient moved during surgery. This was a huge mistake because the trial results were the opposite of what we expected. However, the failed trial created a clear distinction between what we thought we needed to measure and what was being measured. It was a big ‘Aha!’ moment.”

The team did more pharmacology work and learned that what was being measured is not the analgesic effect of pain drugs but rather the consciousness-suppressing effects of hypnotics. They re-designed the trials and this time succeeded, shares Chamoun.

After two near bankruptcies, they introduced the BIS™ monitor to the world, but it still wasn’t an instant success. Chamoun cited two lessons learned. First, the statement “If you build it, they will come,” isn’t always true. Despite the interest and fascination, doctors and hospitals weren’t lining up to buy the product.

Secondly, the market changed considerably from when they first started working on the technology. There was a shift from, “We want to buy innovation at any cost,’ to ‘We want to buy innovation that improves quality of care or reduces cost,” said Chamoun, likening it to current the current climate.

To encourage use among anesthesiologists, the company started another set of collaborations with researchers around the world. They discovered that using the BIS monitor resulted in:

  • Lower costs due to reduced amounts of anesthetic used.
  • Patients waking up sooner with a speedier recovery because they weren’t as groggy. They also required less antiemetics and were more comfortable.
  • A reduction in the risk of intraoperative awareness.

“It’s important to note that we had an amazing team of scientists, engineers, and collaborating anesthesiologists in this country and around the world who helped make the BIS™ Index a reality. I’m very proud of the work we’ve done and our contribution to anesthesia. It was an incredibly gratifying experience, and the innovators in the anesthesia community really helped us figure things out.

“Now BIS technology is available in a majority of hospitals in the U.S., Europe, and Japan, has been used on approximately 100 million patients, and has been the subject of more than 3,500 published articles and abstracts,” says Chamoun. Aspect Medical Systems was sold in 2009 and is now part of Medtronic, where the technology is available today.

After selling Aspect, Mr. Chamoun ran a financial services company before co-founding HDAI in 2016, a seminal year. “That’s when Medicare allowed access to data from 140 million patients from 1999 to the current day, as it refreshes, which is nearly half a trillion records. My vision was to use this information to build an AI platform that can summarize and synthesize information to help clinicians with the burden of electronic health records (EHRs). EHRs have turned doctors and nurses into clerical workers at an enormous cost to the system. We believe AI stands at a crossroads to not only help relieve this burden but to improve many aspects of care delivery.”

These aspirations became a new product known as HealthVision™, a comprehensive AI-powered suite of software and services that drives personalized and proactive care planning, population health improvement, and high-quality network building.

While AI cannot make clinical decisions, it can do all the mechanical work. AI can ingest the medical record, synthesize, and reduce it to one-page that can be read within 30 seconds. The summary is given to every member of the care team, which normalizes knowledge by turning data into insights.

“Anesthesiologists can use HealthVision™ for perioperative optimization and to manage OR staffing based on the risk of the patient,” he explains. “When a patient is admitted, all clinicians can view in real-time the risk for a wide range of adverse events and utilization measures. This granular risk stratification also gives clinicians the ability to design innovative workflows that leverage better care coordination. Of course, anesthesiologists can get there by searching many pages in the record, but as doctors have more patients and less time, AI-synthesized summaries can efficiently highlight areas of focus for the whole care team.”