
Has the cart been placed in front of the horse? The case of AI medical software regulation in developing countries.
Medical software is defined as the use of software for medical purposes. The uptake of medical software in healthcare has increased in line with increased application computation in healthcare delivery. Examples of medical software include software used in bedside monitors, MRIs, PACs, radiation therapy software, infusion pump rate devices, smartphone-based health applications. Etc.
The International Electrotechnical Commission considers medical software as any software system that is developed for the purposed of use in a medical device or itself as a medical device. So, there is a clear linkage between medical devices and medical software. Yet, the understanding of what is medical software and how it should be regulated is unsatisfactory in developing countries. While there has been considerable improvement in the way medical devices are regulated in developing countries, there hasn’t been any serious deliberation of how medical software should be regulated in these contexts.
The emergence of AI software, in particular deep learning, as a resource to aid clinical decision and in some instances automate clinical interpretation has created options for healthcare delivery unlike before. With demonstrated successes in medical imaging interpretation and disease screening , deep learning software has potential for improving access to healthcare delivery in areas where there is limited or no specialised medical workforce. This ability hasn’t escaped the attention of hospital groups, and entrepreneurs alike in developing countries. As low-cost computing and mobile data plans and better processes of data collection get established in developing countries, they also provide a platform for training and deploying AI medical software. Software that can have a significant beneficial impact on patient outcomes. Yet, the frameworks to regulate the use of such software is missing. Without such regulation, the ethical, legal and equity issues that many commentators have identified with unregulated use of AI medical software can lead to unfortunate consequences for the target communities.
Using India as an example of a developing country, we observe both the government and software developers have quickly identified the potential for use of AI in many sectors including healthcare. In recent years, many examples have emerged where technological companies have partnered with private hospital groups to incorporate AI medical software in routine clinical delivery. Also, entrepreneurs with flagship AI medical software products have raised millions of dollars leveraging the buzz around AI. Yet, it isn’t clear in these instances how these applications have received necessary ethical and regulatory approvals and whether there are monitoring frameworks are in place to ensure the safety and privacy of patients.
Relative to many developing countries, the state of regulation of medical devices in India is advanced. With the Medical Devices Rules, 2017 , an improvement on the Drugs and Cosmetics Rule of 1945, there has been streamlining of how medical devices will be approved and regulated in India. These measures have enabled better harmonisation of Indian regulatory standards with medical device regulatory standards in developed countries. Yet, India falls behind countries such as the US and Australia in terms of medical software regulation. Medical software regulation is highly advanced in these countries. In particular, the US has been a world leader in setting out AI medical software regulatory policies. Its Food and Drug Administration (FDA) has laid out a road map as to how AI medical software will be approved and its ongoing use monitored in the US. The purpose of these FDA regulations is not to restrict the entry of AI medical software in the market but rather to facilitate entry of appropriate AI products in the market and to ensure both the interests of the software developers and patients are protected.
As there is significant interest in India regarding the use of AI in healthcare delivery, it becomes all the more important that appropriate arrangements guide the entry and use of AI medical software in the market. To enable this the Central Drugs Standard Control Organisation (which currently acts as the National Regulatory Authority for medical devices in India) can coordinate medical software regulation as other national regulatory authorities in developed countries such as the US FDA and Australian Therapeutic Goods Administration do. Also, customised AI medical software regulatory processes need to be issued similar to what FDA has attempted. Further, it is important that the data acquisition and utilisation process for developing commercial AI medical software is appropriately monitored through relevant regulatory mechanisms. In addition, the government should guide AI medical software developers to incorporate ethical measures in all aspects of their software development and testing. At this stage, these measures aren’t in place, but we continue to see reports of use of AI medical software in various aspects of healthcare delivery. Therefore, is important that the Indian government intervenes to establish appropriate regulatory arrangements not only to protect the interest of the consumer but also ensure the benefits of AI in healthcare are appropriately derived.
While the above narration relates to India, parallels can be found in many other developing countries where the AI medical software is being rolled out without regulatory arrangements to ensure appropriate use. A case of the cart being placed in front of the horse.
Associate Professor Sandeep Reddy is a certified health informatician and Artificial Intelligence (AI) in Healthcare researcher based at the Deakin School of Medicine, Geelong. He has a medical and healthcare management background in addition to having completed AI and machine learning training through several sources. He is currently engaged in research about application of AI in healthcare delivery in addition to developing AI models to treat and manage type 2 diabetes. Also, he has authored several publications about the use of artificial intelligence in Medicine. Further, he is advising several start-ups that are focused on providing AI solutions for healthcare issues. LinkedIn@docsunny50Website