Episode #1: Artificial Intelligence in Clinical Trials has been saved
Episode #1: Artificial Intelligence in Clinical Trials
Life Sciences Connect
Traditional ‘linear and sequential’ clinical trials remain the accepted way to ensure the efficacy and safety of new medicines. However, these traditional trials lack the flexibility and speed required to develop complex new therapies that target smaller and often heterogeneous patient populations.
Unlocking real-world evidence (RWE) using predictive artificial intelligence (AI) models is becoming a business imperative to support novel trial designs, and to identify suitable patients and key investigators to inform site selection. Regulatory bodies have released guidance to encourage biopharma companies to use data-driven strategies to make trials more accessible and participatory.
The COVID-19 pandemic has disrupted hundreds of thousands of ongoing clinical trials around the world. Moreover, there has been a shift in biopharma’s focus to develop vaccines and drugs to combat COVID-19. As the human and economic devastation of this pandemic has become evident, activity promoting the use of advanced technologies in clinical trials has increased.
For our first episode, we explore the use of AI technology in clinical trials and how the COVID-19 pandemic is promoting the use of advanced technologies within the biopharma industry.
This episode is led by our host Karen Taylor. Karen is joined by Fiona Maini, Principal Global Science and Compliance at Medidata, and Fareed Melhem, Head of Acorn AI Labs at Medidata Solutions.
This episode covers:
- How AI technology is currently being used in clinical trials
- How regulators are embracing AI in clinical development
- How regulators are changing their approach in response to COVID-19, including in the acceptance of AI technologies and RWE strategies
Find out more
If you are interested in any of the topics discussed during this episode, please find useful links below:
Stay up to date with Deloitte Centre for Health Solutions
Karen Taylor (00:00:5): Welcome to “Life Sciences Connect”, Deloitte’s Podcast on Life Sciences industry. This series features conversations with leaders from across the Life Sciences ecosystem as well as insights from the critical issues facing their organisations.
Karen Taylor (00:00:28): My name is Karen Taylor and I lead the Deloitte Centre for Health Solutions, an independent research hub supporting Deloitte’s Health Care and Life Sciences industry teams. One of the most fascinating aspect of my job is meeting with innovative companies to explore their solutions and the impact that they could have on everyday lives. The focus of this podcast is on the impact of Artificial Intelligence (AI) on clinical trials and explore some of the issues that we raised in our February 2020 report Intelligent Clinical Trials. As we know traditional ‘linear and sequential’ clinical trials have been an accepted way to ensure the efficiency, efficacy and safety of new medicines. However, these traditional trials take between 10 and 12 years to bring a new drug to market, and lack the flexibility and speed required to develop complex next generation therapies and target smaller and often more heterogenous populations. Key findings include how AI, particularly machine learning and natural language processing, can reduce clinical trials cycle times while reducing the cost and improving the productivity and outcomes of clinical development.
For our conversations today, I am joined by Fiona Maini. Fiona is a Global Compliance and Strategy Principal at Medidata with over 20 years of experience in delivering key strategical and operational excellence programs for both pharma and regulatory clients. Fiona currently chairs the Association of Clinical Research Organisations (ACRO) Decentralised Clinical Trials Working party. Together with Fiona, we have Fareed Melhem who leads the Acorn AI Labs at Medidata. It is Medidata’s data science innovation lab, focused on using advanced analytics to answer the most important questions in Life Sciences R&D and the commercialisation of R&D. Fareed has over 10 years of experience delivering high quality, transformational projects and he is an expert on artificial intelligence applications, such as in R&D and clinical strategy and trial design. Before we get into our discussion, I am going to invite both of my guests to say a little bit more about themselves. So, Fiona can I just turn to you please.
Fiona Maini (00:02:36): Hi. As you mentioned I work for Medidata and chair the ACRO party on decentralised trials and also work within the AI alliance with the EU. I work very much with trade associations and regulators within this ecosystem and really looking at bringing technologies to the regulators, bringing them insights and intelligence on some of the future innovations or current and future innovations that are happening within the clinical research domain.
Karen Taylor (00:03:09): Fareed, would you like to introduce yourself.
Fareed Melhem (00:03:11): Absolutely. Thanks for having me Karen. I am Fareed Melhem. I lead Acorn AI labs at Medidata. Our goal is really to unlock the power of deep data from the 20,000 clinical trials that are run on the Medidata platform, looking for ways to accelerate development programs and also to make trials easier for patients. Also relevant for today’s conversation, my team is actually leading Medidata’s analytics around the impact of COVID-19 on clinical trials.
Karen Taylor (00:03:40): Thank you Fareed and Fiona. Let’s enter into some discussion around some key issues and challenges that are both impacting the Life Sciences industry generally, but also which have been accelerated as a result of the COVID-19 pandemic. First of all, I just like to turn your attention to what we have seen happening more generally and as we note in our report which was published in February 2020, just before the full impact of COVID-19 started to be seen. We note in our report that unlocking real world evidence using predictive artificial intelligence models is becoming a business imperative to support novel trial designs and to identify suitable patients and key investigators to inform site selection. Fareed, what are you seeing in relation to the use of AI currently in clinical trials and could you perhaps explain some of the use cases and patient recruitment approaches that you are using?
Fareed Melhem (00:04:41): Absolutely. We are seeing applications across the development life cycle of AI machine learning starting with trial designs using data both from clinical trials as well as real world data to better access protocol feasibility, define end points and define patient populations. An example, we are doing some work putting historic clinical trial data side-by-side with real world data to better profile patients for clinical trials. We know that patients in clinical trials don’t look like patients in the real world in many ways and so understanding things like background AE rates side-by-side can be quite helpful in trial design. From design we can move into planning and think about country and site selections and the applications of AI there. Safe performances is driven by a range of factors; experience in disease, size of clinical staff, competing trials at the site and the area and complexity of the trial design itself. It is hard for a person to take all these factors into account at once, but a machine can help inform those decisions and give the clinical team the information they need to choose better sites. This is extremely important especially in some of the more crowded disease areas we are seeing where there is a lot of competition for established sites and trying to identify some of those new investigators and new sites, it is going to be critical to continue to enrol these types of trials. Then you move into study conduct. There are applications both in tracking and forecasting of trials to better plan around closeout dates and also in terms of things like central monitoring and risk based monitoring to ensure data quality throughout the trial. The last piece I want to hit is flipping it to the patient angle and actually thinking about the use of AI to better understand the patients themselves and their experience in trials. Whether that is assessing the problem of protocols in patients or building models to predict patient drop-outs and risk foreign patients of discontinuation to help inform support programs. There is an entire patient angle to this as well as trial acceleration angle.
Karen Taylor (00:07:04): So just on that patient angle, how specifically can AI be used to make clinical trials easier for patients?
Fareed Melhem (00:07:14): A couple of ways, one, we can use AI to make trials more inclusive for patients. Meaning identifying sites and investigators who treat patients who may traditionally not be included in trials either because of where they live or because of their access to health care systems. There are a lot of untapped sites out there right now that we can start to profile with AI. The other area is around thinking about protocols from a patient angle. We put a lot of burden on patients in the way that we design protocols today and often a lot of unnecessary procedures and visits. We can use both AI but also virtualisation to help make trials easier for patients to get through.
Karen Taylor (00:08:06): Thanks Fareed. Our report also highlighted the fact that regulatory bodies are themselves trying to encourage the adoption of technologies. They have released guidance to encourage biopharma companies to use data-driven strategies to make trials more accessible and participatory and the importance of early engagement between companies and regulators is something that we have highlighted not just in the recent report, but we have highlighted that in a report we also did on the future of regulation in Life Sciences. So, Fiona, from your perspective and from your access to regulators, how do you see regulators responding to increased use of AI technologies in pharma R&D?
Fiona Maini (00:08:47): I think it is very much recognised by regulators that AI is helping to solve some of the world’s biggest challenges not just in our industry but across industries from treating chronic diseases, reducing fatality rates and traffic accidents to fighting crime, climate change etc. Specific to our Life Sciences end of domain we work in AI technologies of creating, as Fareed mentioned, opportunity for improving clinical trial processes by reducing cycle times, costs and improving patient experience. Pretty much most industries, not necessary the Life Sciences, but traditional regulatory frameworks are not really typically well suited to support the fast pace of development. Regulators can often be challenged to keep pace of scientific and technological innovations. But as science and technology advance and bring potential new treatments and diagnostic tools, regulatory science is advancing. In this way, we see that the regulators are really recognising the need to accelerate their understanding and develop an adaptive and flexible regulatory frameworks. We have seen a lot of activity within the AI space and the use of real world evidence strategies, for example the FDA. Obviously also the MHRA have a cross-GxP AI initiative going on. Also, the European Medicines Agency have a regulatory strategy which firmly lays out their approach to artificial intelligence and the use of big data and other technologies as well. So, we are seeing a lot of exciting change within regulatory frameworks that is either inflight or being executed on or being implemented. So, a very exciting time I think for this topic.
Karen Taylor (00:10:41): As we have all seen, the COVID-19 pandemic has had significant human and economic consequences, including disrupting hundreds of ongoing clinical trials around the world. Moreover, there has been a shift in biopharma’s focus to develop vaccines and other therapeutics to combat COVID-19. There is an increase in activity promoting the use of advanced technologies in clinical trials. Question for Fareed; what changes have you seen adopted in clinical trials as the industry responds to the COVID-19 pandemic?
Fareed Melhem (00:11:12): First of all, you are absolutely right, the impact has been immense. We have been doing an analysis of roughly four thousand trials currently running on our platform and we have seen upwards of 80% drop in new subjects enrolment and, those that are enrolled, we are seeing a 20% drop in patient visits on average. It is causing sponsors to rethink current trials as well as new trials that were supposed to start up. In terms of what they are actually doing, they adapt very quickly along the few dimensions. First of all, just understanding what is going on and the evolving situation has been a major initiative of sponsors and CROs. Things are changing so quickly that there is need for real time data to understand what is happening on the ground, which sites and countries are viable, which ones aren’t and start to plan response around it. As we look forward, we are also trying to think about what does recovery look like and can we start to track disease progression and layer in clinical trial strategies and think about what is our strategy for restarting studies as we get through at least the first wave of this pandemic. The other area that they are thinking about is the trial design itself and having enabled better data capture. There is a big component of this study focussed on virtualisation again making it easier for them to get patient data and not forcing them to go to sites - whether that is e-consent, eCOA or Telehealth visits. The other piece we have seen a growing interest is Synthetic Control Arms; using historic clinical trial data or real world data to round out patient populations where out-patient visits may be at risk. The last piece is probably around maintaining the data quality that we were just talking about. Monitors aren’t able to get to sites and so we do need to move to risk-based central monitoring – which by the way the FDA included as a part of their guidance – to better understand missed visits, data quality as well as missing data.
Karen Taylor (00:13:26): Thank you. Fiona, what are the main changes you have seen regulators taking in responding to the COVID-19 pandemic?
Fiona Maini (00:13:34): I have been working in this industry for about 22 years now and I have never seen such an accelerated action in the development of new regulatory guidance on such a collaborative scale as I have seen in the last month in response to COVID. Clearly it was only in March when the World Health Organisation (WHO) declared a pandemic and regulators in health organisations convened virtually, holding global regulatory workshops and meetings to discuss the pandemic and clearly as a result a range of guidances, reponses and stakeholders running trials is released. Obviously at Medidata we have a dedicated team that tracks and responds to any regulatory laws pertaining to clinical trials. Clearly, our analysis of the COVID-19 guidances was no different to any other. In general, we have reviewed all the guidelines and we obviously see that the regulatory guidelines are pragmatic, flexible and calling for where necessary for alternative strategies to be leveraged and implementing the support for ongoing trials and new trials. There is a degree of alignment between the regulators but there are nuances between the countries. So, it is important to consider each somewhat independently. In general, the changes have been, or the updates have been clearly around eight key areas. Clearly the patient safety is the primary concern here. Also, situation of the safety of the site staff and others working operationally on clinical trials. Other areas where there are common themes across the guidances is around risk assessment – doing risk assessment of your trials, deciding whether to halt or continue the trials. There is guidance on protocols amendments and guidance on remote visits vs on-sites. Clinical trials are being halted but also where they are not halted remote visits are being made possible. Centralising remote monitoring is also a topic that is very common. Also, getting investigational medicinal products to a patient if they are still on a trial from home. In summary, those are some of the key things. There is really a key shift to a different mode of operation with virtual visits and interactions being remote.
Fareed Melhem (00:16:07): Can I build on something you said Fiona. It has been really encouraging to see the industry broadly come together in a number of ways around COVID-19 and the response. Regulators, sponsors and other players in the industry working together to adjust the regulatory side. We have seen sponsors come together and share precompetitive data around COVID-19 to help drive treatments. We have also seen a large portion of the real world data and AI community come together again to pull data from the real world and enable it for academics, investigators and sponsors to really study the disease. Internally at Medidata, we have also seen some of these COVID-19 studies start up faster than any studies we have ever seen before frankly. All of that is inspiring, to see everyone come together but also encouraging that there are better ways to do some of these things. Unfortunately, it took a pandemic to unveil some of the them but there are a lot of lessons that will come out of these.
Karen Taylor (00:17:23): So, Fareed looking ahead at the next 18 months how do you expect AI to support different ways of working?
Fareed Melhem (00:17:31): I do think that we will take a lot of lessons from the COVID-19 experience both in terms of AI as well as digitisations of trials. So, the use of many of these patient friendly virtualisation techniques – e-Consent, eCOA and Telehealth – I expect we will learn a lot and I hope that we continue to push on those pieces. On the AI side, I think that we are enabling much more agile decision making as part of this pandemic but also it is an area that we have been pushing forward for a long time. You can design a perfect trial, pick the perfect sites but six or twelve months later when you start up everything is changed, and it doesn’t have to be a pandemic. It can be three new studies have started up and there have been two new drug approvals. Moving to a much more agile approach to trial management and development programs is going to be critical going forward. Data and AI can enable a lot of that.
Karen Taylor (00:18:45): Fiona, for regulators some of the initiatives have been either temporary or emergency responses. So, what do you expect to happen in the next 18 months or so in relation to regulatory approach to clinical trials?
Fiona Maini (00:19:01): I think obviously the guidances that the regulators have put out they have called them temporary – especially the EMA’s last guidance (in the last ten days) there was in bold that “This guidance is only for the period of the pandemic”. However, it almost feels like that there is a world’s big experiment that has been going on or a big pilot if we are able to leverage these new technologies and new processes and it has been working for the patients why would you essentially want to go backward and not forward. I think there is an opportunity for industry to come together and actually access what are the positives that have come out of this awful COVID humanitarian situation, what can we drive as something positive and implement going forward. So, I think COVID-19 has created that real sense of urgency in enforcing change and hopefully some of the changes will stick. So, I think industry and regulators need to come together at some point to discuss that. At the same time International Conference on Harmonisation (ICH) have been rewriting the ICH E8 document “Principles of Clinical Trials” and essentially call it a modernisation of clinical trials. They are also working on ICH E6 “Procedures for Clinical Trials” and they call that a renovation. I am hoping that through this period we might be able to, while revising those documents, maybe build on some of the experiences that we have had during this period and accelerated advancements in getting trials up faster, moving faster or progressing faster and actually leveraging technologies to help enable accelerated drug development.
Karen Taylor (00:21:04): Thank you. I am going to move to our final question and that is for both of you in turn. What one piece of advice would you like to give to our listeners about designing and running clinical trials in the future? – that’s mostly aimed at Fareed. And Fiona I will come back to you on the same piece of advice on engaging with the regulators in the future.
Fareed Melhem (00:21:28): We are going to be in a time of uncertainty for the foreseeable future. Mapping out what the path of this pandemic will be is really a crystal ball exercise. So, I think at least for the next 18 months to two years we are going to have to think about how to design trials in a flexible way and mitigate some of the risks, think about some of the things like broader country footprints to reduce the risk of a spike in any given country. Including Synthetic Control Arms as part of trial designs to reduce the need for enrolment but also to act as an insurance policy. Continue to think about ways to virtualise pieces of trials to make it easier for patients if they can’t get to sites. All of those have to be part of the normal way of thinking now at least for the next couple of years with the pandemic and really should be part of the normal way of thinking going forward anyway.
Karen Taylor (00:22:32): Thank you. And Fiona?
Fiona Maini (00:22:35): I think we have this new collaborative era that the past few years has unveiled where there is way more collaboration and as I mentioned the regulators have these innovation offices where actually it is an opportunity to go and take new innovations and new technologies and describe them and demo these technologies, and have an ongoing dialogue with the regulators. So, I would say to embrace these innovation offices, there seems to be an open door at least in the past few months we have had meetings with the FDA, the MHRA and also the Irish authorities to basically discuss some of the new innovations and technologies. I would say take advantage of the innovation offices and have early dialogues with regulators. They seem to be very collaborative and provide feedback on different areas of innovation. So, it is a new era for collaboration taking advantage of early dialogues and bringing forward new innovations.
Karen Taylor (00:23:46): I would like to thank you both for joining me today and for being so open and frank in sharing your responses and observations and also thank our listeners for accessing this podcast. Join us next month for the next podcast in our series.