Medical AI Progresses by Leaps and Bounds

Taiwanese companies are aiming to become the next disruptive innovators in a rapidly growing market.

In a Northern German living room equipped with rustic oak furniture, this author’s mother holds her smartphone as if it were a handheld makeup mirror for around a minute. She is then instantly relieved to find that her key vital signs are all within limits that do not require the 86-year-old to get on the phone to make an urgent appointment with her general practitioner.  

Since equipping her smartphone with the FH Vitals App supplied by Hsinchu-based FaceHeart Corp., she has been measuring her heart rate, blood pressure, oxygen saturation, respiration rate, and stress index twice daily with a 60-second selfie video. Previously, she had frequently skipped her prescribed twice-daily blood pressure checks.  

“The conventional procedure of placing the cuff and the tubing of the blood pressure machine over my arm was a little bit cumbersome, but my smartphone is now easier to use,” she says. “The easier the procedure gets, the less nervous I am, which in itself already helps a good deal in keeping my blood pressure low.” 

Taiwan has long been a model for the implementation and use of new technologies, and Taiwanese companies like FaceHeart are forerunners when it comes to Artificial Intelligence (AI) in digital healthcare, a sector predicted by data compiler Statista to reach US$170.2 billion in revenue in 2023 and to arrive at US$256.3 billion by 2027. AI uses massive computing power to collect and analyze huge amounts of data from past events, which is then used to predict future events and choose the best course of action to take in response. 

In the case of FaceHeart’s main product, FH Vitals, the AI was trained by recording thousands of patients’ vital signs while a camera captured minimal changes in the coloration of their faces. The technology is based on rPPG, short for Remote Photoplethysmography, a video-based technology used in smartwatches and fingertip pulse oximeters. Since the company’s founding in 2019, FaceHeart has doubled down on its clinical studies to achieve an algorithm that is robust enough to turn a smartphone into a reliable medical device.  

Erik Ramp, senior director and general manager for Taiwan at Edwards
Lifesciences, says early alerts from AI-driven monitoring can prevent complications and organ damage.

FaceHeart hopes for clearance as an SaMD – short for Software as a Medical Device – a regulatory category created by the U.S. Food and Drug Administration (FDA) during the pandemic, when contactless remote diagnostics abruptly became an important medical topic. FaceHeart expects to gain FDA approval by the second half of this year, which would make it a first mover in the U.S. market.  

“We are trying to be innovators in this sector, riding on a surge towards remote monitoring that is driven by factors ranging from Covid-19 to labor shortages in medical institutions,” says Morris Chung, FaceHeart’s co-founder and vice president of marketing. “It is a quickly growing market in which we can supply companies for their own corporate wellness programs, telemedicine providers, elderly care homes, individual doctors, and smart hospitals.”  

Another company illustrating the pioneering ability of Taiwanese companies to spot promising healthcare niches that could benefit from AI-driven solutions is Taipei City-based Heroic Faith Medical Science. The company has developed an AI-powered continuous breathing sound monitoring solution that can be used under procedural sedation (non-intubated anesthesia, or minimal to moderate sedation), a process that is becoming more and more widely used. 

The device, sold under the brand name Airmod, looks like a lightweight stethoscope patch connected to a phone with a palm-sized converter. It uses deep learning and algorithms trained on more than 1,000 patients to analyze sounds and send alerts for abnormal breathing situations, even in very noisy environments, by utilizing active noise cancellation. This feature is a major selling point for dentists, whose drills emit noises making it nearly impossible for anesthetists to monitor patients with conventional stethoscopes.   

“On the one hand, there is a trend of doing dental and plastic surgeries with the patient sedated, and on the other hand, there is a risk of unexpected airway obstruction, apnea, or hypoxemia due to breathing accidents during anesthesia,” says Cindy Huang, Heroic Faith’s marketing representative.  

“Airmod makes it easier to avoid unnecessary hypoxia that leads to serious consequences like emergency intubation, hypoxia complication, or even death, and clinics often do not have sufficient hands to monitor the patient’s status continuously.” 

Airmod gained Taiwan FDA approval in 2021 and was approved by the U.S. FDA in March 2023. Currently more than 20 medical centers are using the product, and a range of U.S. hospitals – including the Cleveland Clinic, Stanford University Medical Center, and UCLA Medical Center – have already partnered with Heroic Faith for use in clinical trials. Huang says business is expected to pick up markedly this year as clinical studies progress and positive user feedback is collected.  

Critical solutions 

Although Taiwan’s relatively small population means limited demand, medical AI in Taiwan consists not only of Taiwanese companies seeking to conquer overseas markets but also of foreign companies distributing their technologies to customers on the island. A leading example is Edwards Lifesciences, an American medical technology company specializing in replacement and repair of heart valves and hemodynamic monitoring.  

The company supplies an AI-driven bedside monitoring system involving a set of disposable sensors that can be used non-invasively or connected to the arterial lines of high-risk patients, such as those at risk for heavy blood loss after open thorax surgery. The technology essentially measures tiny blood pressure changes, with an AI that has been trained on historical data establishing the likelihood of a given change in the cardiac waveform being followed by a particular undesired event, such as a dangerous drop in blood pressure. 

“This can alert the physician up to 10 minutes before that the event may strike and thereby give the physician some lead time to administer fluids and medications to stabilize the patient,” says Erik Ramp, senior director and general manager for Taiwan at Edwards Lifesciences. He adds that early alerts can prevent complications and damage to vital organs. When blood pressure drops to a dangerously low level, the kidneys are often deprived of oxygen or nutrients and be damaged in what is called Acute Kidney Injury (AKI). 

This form of AI-driven support in medical decision-making is becoming widely recognized by Taiwan’s anesthesiologists and critical care physicians for its value in preventing perioperative complications (complications around the time of surgery) in patients. For example, AKI can help patients on dialysis for end-stage renal disease, which constitutes an enormous burden on Taiwan’s national healthcare. Edwards’ AI system is under National Health Insurance review, with the company hoping for a positive decision later this year.  

“Because of the complexity of AI and the inability of critically ill patients to be involved in the decision about their hemodynamic monitoring, we believe reimbursement could pave the way for more high-risk patients to receive this type of monitoring,” Ramp says. He notes that AI applications in medicine are very promising but are still far from becoming the “gold standard” in patient care.    

AI can also help hospitals manage their financial burdens. One company applying such solutions is Taipei-based AESOP Technology. Its two solutions – RxPrime and DxPrime – counter medication and billing errors, respectively. 

The AI was trained on 3.2 billion electronic medical records from Taiwan and the U.S. RxPrime uses this data to flag instances in which doctors mistakenly prescribe the wrong medication. DxPrime automates the work of medical coders, whose job is to review the patient’s file, and translates file data into universal codes required to arrange treatment cost disbursements from insurance companies. The financial implications of this solution are enormous, given that a typical mid-sized hospital in the U.S. needs to employ dozens of coders.  

RxPrime and DxPrime sit atop Epic EMR, the electronic medical record software system commonly used by U.S. hospitals, meaning it can be easily adapted and deployed. Revenue is generated under schemes in which hospitals pay a share of the money saved by the reduction in medication and billing errors. Currently, AESOP’s primary revenue comes from Taiwan, but the company is also conducting clinical trials with several hospitals in the U.S. and is actively expanding into the U.S. market. 

“For a middle-sized hospital in Taiwan with 600 beds, we can save them up to NT$6 million per year,” says Jim Long, CEO of AESOP Technology, noting that the systems are typically first employed in individual departments before the use is expanded to the whole hospital. “Automation of costly administrative tasks is clearly a key trend for the healthcare sector because fewer people are willing to work in healthcare as they are put off by the higher workload and the risks affiliated with the pandemic,” he says.    

Robots in the operating theater 

Today, medical AI does more than manage clinics’ digital patient files, assist doctors in providing a diagnosis and treatment plan, and guide our at-home follow-up care. It also drives robotic assistants, which are already being used in a wide variety of surgeries and in hospital logistics for the distribution of medications. Robotics provides higher precision, better visualization, and a minimally invasive approach to surgery. 

The Department of Obstetrics and Gynecology at Taipei Medical University performs robot-assisted surgery with equipment from U.S.-based Intuitive Surgical. In addition to the innovative equipment designed to improve surgical quality, doctors at the department note that in their experience, patients that undergo robotic surgery can more quickly return to normal life. 

“Uterine and endometrial cancer surgical patients used to have to stay in a hospital for up to several days,” says Dr. Lai Hung-cheng, the department’s director and one of the first medical professionals in Taiwan to practice robot-assisted surgery. “But now, a patient of mine can fly into Taipei from San Francisco, undergo surgery on a Wednesday, and be discharged the next day. She can then fly to Hong Kong for business on Saturday and be back in her San Francisco office on Monday.” 

But Lai notes that despite the many benefits of robot-assisted surgery, Taiwan has been slow to adopt the technology. One of the reasons he cites for this slow uptake is that surgeons in Taiwan are highly skilled in laparoscopic surgeries, while robot-assisted surgeries are mainly used when traditional laparoscopic operations are impossible. A more significant reason is the initial cost of investment in medical devices.  

“I have a simple calculation that exemplifies the cost-effectiveness of robotic surgery long-term,” says Dr. Lai. “By just switching all regular hysterectomies to minimally invasive surgeries for the appropriate patients using robot-assisted tools, you can save about 38,000 hospital-stay operation days in five years. That means you can increase healthcare capacity by over 20 beds per day.”  

Looking forward, Lai says he will continue pushing for the increased use of robot-assisted surgery. He points to Denmark, which has adopted robotic surgeries as a standard for endometrial and cervical cancers, as an example of prospective benefits for Taiwan. After more than 10 years of use, Denmark reports patients had a reduced risk of severe complications compared to open surgery.  

“The more complicated the situation is, the more advantageous robotic surgery is,” Lai says. “If more surgeons start using robot-assisted surgery, the outcome for gynecological cancer patients in Taiwan could drastically improve.”