From Drug discovery to delivery: AI's impact on Pharma and Precision Medicine

Contributors:  

Susanne Munksted (Chief Precision Medicine Officer), Scott Gamester (VP Platform and Data), Scott Morrison (Senior Director Data Science), Brien Mullins (Senior Director of Engineering) 

Key takeaways 

• The impressive potential of Artificial Intelligence (AI) is impacting the pharmaceutical industry at different levels: drug discovery and development, clinical research and commercialization,  
• We expect to see a boost in the number of partnerships between pharma and AI companies as well as a significant increase in AI and data-related positions in pharma companies, 
• Diaceutics has a unique real-world healthcare database and is actively implementing AI solutions to allow pharma and biotech to harness the full potential of the data. As such, we support clients in understanding the diagnostic landscape and gaining access to real-world data. Our activities eventually contribute to getting the right patients faster on the right treatment, 
• To conclude, AI is not the magic bullet for fixing the broken system in healthcare and precision medicine. However, as AI and its numerous applications mature, it will offer pharma the opportunity to optimize targeted therapies and associated biomarker testing development and adoption, eventually fully enabling precision medicine. 

Introduction  

Assistant, augmented, autonomous, artificial...intelligence (AI). No matter what we call it there is one reality: AI is not another hype; it is already integrated in our daily lives and activities. And although it is not a new concept - with multiple companies trying for the last decade to apply it to healthcare with limited success - it was only with the launch of ChatGPT at the end of 2022 that the sleeping beauty has finally fully awakened. Whether the distant future will look like “Star Trek” or  “Terminator” is still unknown… but for the near future, there is more impact to come in the AI field, especially in the healthcare sector.  

In particular, let’s take a look at precision medicine (PM) which aims to treat patients based on their unique genetic makeup and medical history. As big data is the foundation of PM, AI is uniquely poised to be a natural partner with the goal to revolutionize the way diseases are diagnosed and treated. As such numerous applications have arisen, especially in the field of digital pathology, which can be used to pre-screen patients (see our Better Testing, Better Treatment podcast ep.2: AI in Breast Cancer: From Labs to Clinic). However, the implementation of AI in the clinical PM routine presents unique challenges.  

Here we will focus on AI in healthcare and its hopes and fears in the next few years. We will also discuss how we at Diaceutics are integrating and leveraging AI. While the technical aspects of AI are out of scope, we will be discussing the overarching AI representing all subsets and branches e.g. natural language processing, machine learning, deep learning, etc.  

AI in the pharma industry 
 
AI's multifaceted contributions are reshaping the landscape of healthcare, and as such, impact different key stakeholders. For instance, in the world of diagnostics, AI is a powerful tool for interpreting large datasets such as image-based radiology data, genomic, and pathology data. Hence, the use of AI in radiology and digital pathology has increased substantially, improving inter- and intra-variability. In addition, over the last months, the promise of AI on improving drug discovery has started to materialize into more direct clinical impact, with several biotech's launching clinical trials enhanced or driven by an AI system. Recently, the Precision Cancer Consortium set up a collaboration with Massive Bio to use their AI analytics and enhanced clinical trial matching based on patient’s NGS data¹, with the goal of addressing one of pharma’s big challenges: identifying patients with a specific genetic profile to enrol in clinical trials.  

Given the impressive potential of AI, many pharma giants and biotech are starting to highlight their AI efforts as high priority in their earnings reports. This includes the use of AI tools during manufacturing, drug development and commercialization. Let’s take 2 examples:  

The first one, Sanofi, has embraced AI in various ways and revolutionized its operations, building, acquiring and/or partnering with multiple AI programs from drug discovery and development, to manufacturing, clinical trial optimization and data analysis to support PM and stakeholder engagement. For example, their Omnichannel AI Engine provides field teams with actionable insights on the next best action to reach out to more than 90,000 HCPs through tailor-made engagement plans. More interestingly, stepping further into a comprehensive digital transformation “at scale”, Sanofi introduced an AI application (plai) in June 2023. This app facilitates instantaneous and responsive data interactions and provides valuable insights spanning all of Sanofi's operations. The aim is to empower decision-makers within Sanofi to make well-informed and faster choices across the value chain. 

Similarly, Moderna has embedded AI at multiple steps of its operations, including clinical trial recruitment, PM vaccine development and sales strategy empowerment. In particular, AI has helped Moderna in designing mRNA sequences for a COVID-19 vaccine, and in speeding up its development and production (shifting from manually manufacturing 30 mRNAs vaccines each month to around 1,000 vaccines with the support of AI algorithms and robotic automation). The outcome being that not only did AI support Moderna to be among the first companies to release an effective, life-changing Covid vaccine and increase its revenues, but also to gain trust and brand visibility among patients, HCPs and investors. 

Diaceutics’ opinion on future impact of AI in PM 

So what are these 2 examples, among many others, telling us? As many blockbuster drugs approach their patent expirations and new competitors (including tech companies) enter the space with innovative approaches, pharma are looking for new approaches to cut costs, increase success, speed up the time to bring innovative drugs to market and to improve stakeholder adoption. As such, business as usual is no longer a viable option to thrive in this 4th industrial revolution; Instead, disruptive solutions that fit within the larger internal and external ecosystems (beyond the pill) will be needed. 

Expectations in the short term: 

1. More pharma partnerships, deals and acquisitions with tech-savvy AI start-ups. Takeda was among the first pharma to drive this by buying an experimental psoriasis drug developed with the support of AI for $4 billion from Nimbus. 

2. More data scientists added to the team, new positions as chief data, generative biology, AI ethics and/or AI officers to oversee and drive AI optimal use . 

3. Optimized stakeholder engagement. Although the field is more advanced to support preclinical and clinical drug development, AI will be integrated in all pharma operations. 

We expect pharma companies who have adopted AI today in multiple divisions to be ahead of their more conservative competitors, with cross-functional teams using this extra support to drive development, commercialization and sales of innovative therapies. Beyond drugs, we also expect diagnostic testing to become easier to analyze and more sophisticated in predicting responses to treatments, and therefore more accessible and better integrated within the patient journey. As such, AI will support pharma to discover efficient targeted therapies linked to more accurate biomarkers, leading to faster and more successful drugs and companion diagnostics being brought to market. The estimation is that over the next decade, AI drug discovery could lead to an additional 50 novel therapies worth $50 billion in sales². In addition, leveraging AI to better understand patient PM journeys will lead to more relevant, timely and effective engagement with stakeholders, eventually improving drug matching to provide the most appropriate therapies for patients. 

How Diaceutics is leveraging AI solutions to improve precision medicine

Diaceutics’ unique diagnostic and healthcare data repository is a multisource database that consists of commercial and Medicare claims, EHR, prescription and real-time laboratory data covering >400 million lives. However, data coming from all these sources is sometimes unstructured, non-uniformed, biased and/or missing details, preventing an accurate, comprehensive analysis. 

To address these challenges, Diaceutics has embraced multiple AI solutions. For example, leveraging proprietary machine learning techniques and data science have allowed for more accurate and effective labeling, structuring and standardization of data at scale, as well as an analysis of patterns to fill gaps in diagnosis, biomarkers, test methods and detected alterations. As such, Diaceutics, powered by our DXRX platform, can mine valuable information across its entire dataset, and more precisely define, segment and customize the patient cohort of interest. This is leading to faster and more accurate identification of HCPs caring for patients that would benefit from a specific therapy, and therefore more timely and targeted engagement – even before HCPs receive lab results - to ensure no patient is left behind ( with our daily and weekly Signal product). It has also supported gathering unique insights from real-world data for better actionable risk-mitigation strategies (e.g. lab billing practices and reimbursement challenges). 

Diaceutics has also been supporting pharma, biotech and diagnostic companies in understanding the market landscape of new diagnostic AI solutions, providing consultancy to drive successful commercialization and increasing awareness and education on the application of AI in healthcare (see our podcast on digital pathology in breast cancer). 

Diaceutics is actively working to develop and implement AI Solutions to allow clients to harness the full potential of data by quickly analyzing and generating insights, identifying patterns and predicting future trends. We can also support in defining the next best action based on data, including commercialization efforts and targeted HCP outreach. 

Final thoughts 

Not everything that shines is gold… AI is a magic bullet for fixing the broken system in healthcare in general or PM in particular; if not planned well, it could lead to more frustration (internal, external), misinformation, mistrust and patient harm. Sometimes even to losses that cannot be easily recovered. However, the field has not yet started walking to even begin to run… it is in fact just still crawling. But as it matures, it will offer the opportunity to optimize targeted therapies and associated biomarker testing development and adoption, thus fully enabling PM.  

For example, we have previously published on the seven clinical practice gaps for the optimal implementation of PM³. AI systems can help by reducing each of these gaps in the patient journey: Clinical decision support tools can be integrated within the clinical workflows to optimize biopsy referral, test ordering and interpretation and treatment decisions. Improved analytical programs and pathology/molecular techniques can enhance accuracy, sensitivity, specificity, reproducibility and turnaround time. AI-driven relevant and targeted HCP/patient education and engagement can improve awareness, adoption, adherence and access to innovative PM therapies. In addition, early data points towards improved accuracy and speed of diagnosis, and reduced healthcare costs, leading to a handful of approvals (e.g. IDx-Dr, the first FDA-approved (2018) and CE-marked autonomous AI-based diagnostic tool). 

A final thought from the famous IT phrase "garbage in, garbage out”: The outcome of any AI system is as good as the data it is fed, a crucial fact especially in healthcare. Unfortunately, our human bias is inherited to machine bias. Other issues widely remain, including the lack of clean, accurate and complete data to ensure reliability, safety and confidentiality, alongside limited storage capabilities for the ever-growing amount of data and shortage of an AI specialized workforce. 

We, at Diaceutics believe in and are working towards data and PM democratization. And so let's end with a call for unity to share data across pharma and healthcare institutions to support decreasing disparities and bias and improving AI output. As such, we now start building towards a ‘Star Trek’ future with a focus on innovation and expanding our potential. 

References 

1. https://massivebio.com/precision-cancer-consortium-teams-up-with-massive-bio-to-utilize-ai-analytics-and-enhance-clinical-trial-matching-for-cancer-patients-worldwide/  

2. Morgan Stanley: https://www.morganstanley.com/ideas/ai-drug-discovery  

3. Sadik, H., Pritchard, D, Keeling, D.-M., Policht, F., Riccelli, P., Stone, G., et al. Impact of Clinical Practice Gaps on the Implementation of Personalized Medicine in Advanced Non–Small-Cell Lung Cancer. JCO Precision Oncology, 2023.