BIS Research recently concluded an extensive and insightful webinar on the “Future of AI in Medical Diagnosis with Emerging Technologies” discussing the prospects of the use of artificial intelligence in the field of medical diagnosis and how it is revolutionizing the overall healthcare industry.
The webinar was hosted by Swati Sood, principal analyst, and Shreya Srinivas, research analyst from the healthcare team at BIS Research. The two healthcare analysts were joined by Mr. Paul Fletcher-Dyer, AI Compliance Director and DPO for Cognetivity Ltd.
Some very critical questions were raised during the session by the attendees, which were duly answered by the panel of speakers.
Here’s an excerpt from the QnA that took place during the webinar:
Q. What do you think are some of the key trends being witnessed in AI in the medical diagnosis space?
A. AI-enabled software solutions are much more prevalent, accounting for nearly 80% of the type of AI-enabled medical devices available in the market currently. However, the integration of AI within hardware systems is a trend that will be witnessed in the coming years.
Even if AI has started as an algorithm, as software is put on a server and connected to a network, it gradually transforms into a device itself. For instance, I believe some AI algorithms will be built inside mammography systems to help the technician make decisions even before the doctor reads the mammogram.
Moreover, AI is expected to play a key role in areas such as predictive analysis, where it is not possible for human beings to always predict the disease prognosis accurately due to the presence of too many data points.
Q. You mentioned that AI chatbots could converse with patients, conduct a preliminary diagnosis, and even refer the patient to the right healthcare professional. What are some of the challenges in employing such chatbots for diagnostic purposes?
A. Yes, while chatbots can be useful in healthcare, they come with their own set of challenges, the biggest of them being an incorrect diagnosis. This could be due to its lack of ability to consider all the factors in the patient’s medical history. Another challenge is with regard to patient acceptance. This is because many patients may prefer face-to-face interaction with a real doctor rather than a bot to diagnose their condition.
Q. You talked about the ability of AI to reduce the burden on the healthcare system and shorten waiting times. Are there any other key opportunities that AI holds in healthcare?
A. Yes, the applications of AI across a wide range of clinical areas present opportunities for further research and development. While a vast majority of AI-enabled medical devices have been approved for radiology, followed by cardiology, AI is being increasingly leveraged for other clinical areas as well.
Companies are venturing out to develop AI-enabled medical devices for other niche applications such as dental, obstetrics, and orthopedic applications. For example, Dentsply Sirona is the only company that has received FDA clearance for its software solution for dental applications. Similarly, Vitrolife A/S is the only company that has received FDA clearance for software for gynecological applications. Such clinical areas hold immense opportunities for market players to venture into and gain market share.
Q. Can AI-enabled medical devices be regulated the same way as other medical devices?
A. Technically, at the moment, the only way AI-enabled medical devices can be approved on the market is in comparison to other medical devices. So it's under the same regulations as the Medical Device Directive or the Medical Device Regulation or the FDA equivalent of those. It's the only way to actually prove those AI medical devices. The issue that arises is that a lot of the actual requirements for those medical devices are based on physical products. So if you took a normal AI product under the Medical Device Regulation, over half of the rules that are required for medical devices would not actually apply because AI-enabled devices are not physical products. It's the only legal way of actually getting an AI onto the healthcare market at the moment, but it's also not the best way of doing it if that makes sense.
Q. How is the regulatory framework surrounding AI-enabled solutions in emerging countries?
A. If I talk about countries such as Brazil and Mexico, until 2021, no regulatory framework governing the use of AI in healthcare existed in Brazil. The General Personal Data Protection Law was implemented by Brazil in September 2020 and aimed to ensure data protection. While the law does not specify AI, it is the closest available regulatory framework in the country currently.
In Mexico as well, there are no particular regulations concerning AI/ML-enabled digital health devices and their approval for clinical use. However, in May 2018, Mexico launched a national AI strategy, which was a key milestone. So, while initiatives are being undertaken to create AI strategies and enhance cybersecurity, well-defined regulatory frameworks are not yet completely in place in emerging countries.
Q. What are the regulatory frameworks in place for addressing cybersecurity concerns when it comes to sharing medical data?
A. While the lack of available data to the AI system can potentially lead to higher chances of inducing bias, it also raises concerns relating to data sharing among third-party sources. Most countries across the world have laid down strict privacy laws and regulations, which need to be followed to get patient information. For instance, the U.S. has the Health Insurance Portability and Accountability Act (HIPAA) regulation to ensure patient privacy. Similarly, the General Data Protection Regulation (GPDR) in the European Union provides individuals control over their personal data and also processes health information that is highly sensitive. However, these are for the general protection of patient data, and there is a need for more AI-specific cybersecurity laws to be implemented.
Q. Have many countries globally implemented national or local AI strategies specifically aimed at healthcare?
A. Many countries have implemented national and local AI initiatives. For instance, according to data published in OECD.AI, by the end of 2021, the U.S. had the largest number of AI initiatives at 77, followed by 57 initiatives in the U.K. and around 35 in countries such as Germany and France. While this gives us an overall insight into AI initiatives across different areas, healthcare is usually a prime focus of these initiatives, even if it isn’t specific to healthcare only.
Watch the complete webinar below: