From data to decision-making: How AI will transform medical affairs - European Medical Journal

From data to decision-making: How AI will transform medical affairs

7 Mins
EMJ GOLD
AI is becoming less of a buzzword and more of a key consideration for those in the pharmaceutical industry. For medical affairs professionals, the benefits of AI are growing – but how can they be harnessed? 
Interview by Jade Williams

In the dynamic realm of medical affairs, teams are tasked with a dual challenge: adapting to evolving industry needs while maintaining their core mission of educating and engaging healthcare professionals.

At EMJ GOLD, we believe that medical affairs can be a transformative force, and EMJ offers innovative tools to help them succeed along the way, but AI is a road less travelled. To explore its potential, we asked three experts to share their insights on how AI will reshape medical affairs. Where will it have the greatest impact, and what are the potential challenges to watch out for?

Contributors:

  • Carlos Eid, Executive Director, International Medical Affairs, CV & ASCVD, Novartis 
  • Ankur Sharma, Vice President, Head of Medical Affairs, Medical Devices & Digital, Radiology, Bayer  
  • Ludmil Alexandrov, Chief Scientific Officer, io9 

What do you see as the most important current applications of AI in medical affairs?

Carlos: One of the most important applications of AI in medical affairs is data analysis and the generation of actionable insights. AI can process big datasets, including clinical trial data, real-world evidence, medical science liaison notes and scientific literature, uncovering insights that previously took weeks or months to obtain. This capability allows medical affairs teams to lead cross-functional discussions and make more informed (and timely) decisions around strategy and tactics. 

Additionally, AI is increasingly being used in KOL identification and engagement, where it not only helps to identify the most influential experts but also analyses their sentiments and communication patterns. 

Ankur: It’s about large language models (LLMs), right? The role of medical affairs, clinicians and their thought processes – how they analyse data sets and different points and then process them to produce an output that might develop a new product, shorten a development cycle, or engage with external partners or HCPs – is tricky and time-intensive. 

LLMs can change how you analyse and process data, allowing you to do it in a shorter time frame, which is a big deal. Having a tool such as an LLM and using it to query a theoretical idea can help you look at extra data in the space you’re examining. It can help you decide if all this thing will improve patient outcomes or drive innovation for the population. 

Ludmil: AI holds the promise to revolutionise medical affairs by enhancing diagnostic accuracy, personalising treatment plans and improving early detection and screening. 

The most immediate and impactful application of AI will be in diagnostics, particularly cancer diagnostics. By utilising AI to predict disease characteristics directly from histopathological slides, we can eliminate expensive and time-consuming molecular testing. This approach will expedite diagnosis and enable more precise, timely and personalised treatment decisions, significantly improving patient outcomes. 

Are there any major limitations to their implementation, and if so, what advice would you give to companies looking to implement them?

Carlos: There are several limitations that come to mind. Data privacy concerns, integration challenges, inaccuracies, biases and lack of talent/expertise. 

My advice to companies looking to implement them is to invest in robust data security measures and ensure compliance with regulations like GDPR or HIPAA – or partner with providers that have mastered this step and have these measures in place. Second, build the right workforce with the right talents. Make sure you have teams that understand the power of AI and can leverage AI effectively. 

Third, AI in medical affairs is like a high-performance sports car – it’s powerful, but if you fuel it with bad gas (or in this case, bad data), you’re not going anywhere fast – AI’s conclusions could be inaccurate and biased. Therefore, treat your data like gold, invest in clean, representative datasets and keep good human oversight on all AI outputs. Finally, start with small, pilot projects to help integrate AI into your broader medical affairs operations. 

Ankur: The limitations here are significant, especially regarding data access and data rights – these are well-known challenges in healthcare. Data access is difficult because of patient information and rights, and now we’re talking about giving that data to an AI, which introduces even more challenges. There are also ethical considerations. When you feed data into LLMs, what are they going to generate with it? How will they use it? It’s crucial to integrate LLMs into the workflow in a way that carefully considers the data itself. 

Cybersecurity is crucial. How do you protect patient privacy and PHI when using these AI tools? These aspects need to be taken into account when developing a product, or when helping other organisations to implement new AI-enabled healthcare applications, for example to assist radiologists in their daily work with critical and often time-consuming tasks. We need to be aware of these challenges before releasing any AI development tool, and this is something we are taking seriously, and others will need to consider as well. 

The second aspect is getting users to maximise the potential of these tools, which is known as prompt engineering. It’s about how you construct the prompt to get the right answer for what you need. For instance, you could give an LLM 50 papers and ask it to summarise them, and it would do a fine job. But if you don’t specifically ask for any common themes or recurring elements in those papers, you might miss important connections, losing some of the LLM’s value. So, how you build and use LLMs, and the transparency behind them, is crucial. 

Ludmil: Yes, there are major limitations to implementing AI in medical diagnostics. These include data privacy concerns, the need for large and diverse datasets for training, regulatory hurdles and integration challenges with existing healthcare systems. 

My advice to companies looking to implement AI solutions is to prioritise robust data security measures to protect patient privacy, invest in acquiring and curating high-quality datasets and collaborate closely with regulatory bodies to ensure compliance. Additionally, companies should focus on developing interoperable systems that can integrate with current healthcare infrastructure and provide comprehensive training to healthcare professionals to facilitate smooth adoption and utilisation of AI technologies. 

Most importantly, [companies should] build better AI models that can be trained with smaller amounts of data will provide unprecedented advantages. This approach will make AI technology more accessible to institutions with limited data resources, reduce the time and cost associated with data collection and processing  and enable quicker iterations and updates to the models. 

How do you see AI impacting, positively or negatively, the collaboration between medical affairs and marketing departments?

Carlos: AI helps both teams align by providing real-time insights into market trends, latest scientific updates, KOL identification, sentiment analysis and HCP preferences, etc. This fosters data-driven strategies and enhances inter-departmental communication. 

By accessing and analysing these vast amounts of data, we will witness better-targeted marketing campaigns and more accurate scientific communication and data dissemination. This data-driven approach ensures that marketing efforts align with the latest scientific evidence and regulatory requirements, improving the credibility and effectiveness of campaigns. 

However, there could be challenges if AI tools are not fully understood or if there is a lack of alignment between the two departments in how AI is utilised. To keep things harmonious, we need to set clear boundaries and ensure both teams understand AI’s role. AI can make these departments work like a well-oiled machine – as long as they remember to keep the lines of communication open. 

Ankur: Overall, I think the outlook is positive, with a strong opportunity for growth and collaboration between medical affairs, R&D and marketing. However, it’s important to be careful about what you’re asking for and how it’s interpreted. 

You need to consider how much of this process is driven by AI, how much is influenced by LLMs and how much involves real-world human translation of ideas into execution. This applies whether it’s R&D working on a new product or marketing developing a campaign and crafting claims.  

This multidisciplinary cooperation is valuable, as different groups – medical professionals, technical experts, or marketing teams – speak different languages. Without transparent, collaborative communication across all these groups, it becomes a challenge. However, LLMs can help by bringing these elements together, enabling more effective collaboration. 

By focusing on the variations, rather than the broader picture which is already distilled for you, LLMs can help develop better marketing strategies and streamline data analysis. It’s a balance between these AI-generated outputs and human decision-making, where people discuss the differences and nuances. But overall, I think it’s positive. 

Ludmil: AI can positively impact collaboration between medical affairs and marketing by enhancing data sharing and analysis, leading to more aligned strategies and effective, data-driven marketing campaigns.  

However, challenges include ensuring data accuracy, avoiding biases, balancing different departmental objectives and the risk of overselling AI capabilities. To maximise benefits and avoid hype, organisations should promote collaboration, transparency and clear guidelines to align goals, ensure compliance and set realistic expectations about AI’s potential. 

Do you think the wider use of AI in medical affairs will be beneficial or detrimental to HCP engagement, and why?

Carlos: AI will be beneficial to HCP engagement if implemented thoughtfully. AI can offer personalised, relevant and timely information, helping them stay updated with the latest research, treatment options and guidelines, freeing up time for them to focus more on their patients and making the more complex decisions.  

From the pharma/HCP interaction lens, AI can facilitate better communication by providing tailored content that meets the HCP’s specific needs and interests. But here’s the catch: AI can sometimes be a little too perfect. Like that dinner party where the conversation feels a bit… scripted. Sure, the topics are on point, but something’s missing—the warmth, the spontaneity, the human touch. So, while AI can whip up all the right ingredients—data, insights, content—it’s up to us to add the flavour. Think of AI as your sous-chef. It does the prep work, but you’re still the one crafting the dish. And let’s be honest, no one wants to eat a meal that tastes like it was made by a robot. 

Ankur: I think it helps with HCP engagement, if done right. There’s so much information out there, hundreds and thousands of papers, data sets and real-world studies. How do you distil that, ingest it and provide feedback so that [HCPs] can deliver the highest level of care?  

It’s about taking AI tools to help us reduce that manual work, save time and let people provide strong collaboration with their colleagues but also, really importantly, with HCPs. The practice of medicine is evidence-based, right? Having an AI that can facilitate all the evidence and distil it in a way that helps you make a more personalised, evidence-based decision for your patient would be amazing.  

As doctors learn how to use these things, it’s going to be an unbelievable tool. And it’s really a complementary tool, right? It’s just another tool in the bag of physicians to augment their expertise. As professionals, we go to school for a long time to learn a lot of things, and to have this in your pocket that can remind you of things or collect real data in real time is such a huge positive. I can’t wait for it to be adopted on a higher scale. 

Ludmil: The wider use of AI in medical affairs will likely be beneficial to HCP engagement by providing more personalised and timely information, enhancing decision-making and improving patient outcomes. However, it’s crucial to ensure AI tools are user-friendly and transparent to maintain trust and avoid potential disengagement due to perceived complexity or lack of understanding. 

What innovative or unexpected ways do you think AI could be used in medical affairs in the future, and how should the industry prepare for these advances?

Carlos: This field is rapidly evolving and I am not sure how far away the “future” is. I see great use of AI in predictive analytics for clinical trial outcomes, where AI could analyse data from ongoing clinical trials to predict outcomes and suggest modifications in real-time, leading to more efficient and effective trials.  

We will also see AI generating and personalising scientific content in real time and advanced virtual AI assistants that cover all your time-consuming day-to-day tasks. Why not think of increased collaborations with societies and governments in leveraging AI to monitor and improve patient outcomes? To be ready, the industry needs to start preparing now. Invest in AI literacy, infrastructure, data quality, ethical and regulatory measures and encourage a culture where innovation isn’t just welcomed but celebrated. 

Ankur: People are thinking about interactivity with patients and doctors, and how they can get real-time information about what disease they have and what treatment they need, so that those treatment plans can be tailored. 

Think about how many times you call your doctor and you’re like, “Hey, I need to talk to you”. And they’re like, “Okay, I’ll call you back in a day when I have a free moment”. But now, if you have a smart language model that you can query and get some basic responses that your doctor can then look through, you get a real-time back and forth with this LLM in between to facilitate that. 

And then I think it’s going to be interesting to see how LLMs or AI tools can help optimise the design of clinical trials, how you recruit patients and monitor them – how you facilitate that engagement and education of patients in the process. You could have a personalised patient pathway and give them the custom support they need. And that goes back to the earlier example: how do patients and HCPs interact? If you have this LLM that understands the real-time interaction that’s been provided, it can be so powerful for patients to feel acknowledged about the care of their health. 

Ludmil: In the future, AI could be used in innovative and unexpected ways in medical affairs, such as predicting disease outbreaks, optimising clinical trial designs and providing real-time treatment recommendations based on comprehensive data analysis. To prepare for these advances, the industry should invest in robust data infrastructure, ensure interoperability of AI systems, foster interdisciplinary collaboration and prioritise ethical considerations and transparency to build trust and facilitate the seamless integration of AI technologies into healthcare practices.

Rate this content's potential impact on patient outcomes

Average rating / 5. Vote count:

No votes so far! Be the first to rate this content.