With less than 10 percent of new medicines attaining regulatory approval, the value of a potential asset not only depends on its scientific advancements and potential to serve unmet patient needs, but also on positioning in the broader therapeutic market landscape. Effectively gauging where an asset stands in the midst of an ever-changing, complex competitive space is critical to trial sponsors planning asset development and preparing for launch. In some cases, such as biotechs who rely on one or a few assets to drive their businesses, understanding competitive positioning is integral to company success. 

Given the sheer breadth of data available to pharmaceutical and biotech sponsors, traditional manual processes for extracting insights that guide clinical planning can be time-consuming and inefficient, delaying key insights needed to inform decision-makers. 

Solutions driven by artificial intelligence/machine learning (AI/ML) and designed to aggregate millions of real-world data points and quickly deliver meaningful insights can augment traditional secondary research and expert assessment to enable more objective and faster analysis. When aiming to secure a comprehensive view of a specific therapeutic landscape, sponsors can better gauge how successful their strategy will be and steer their investment efforts accordingly.

In our discussion below, we break down how AI/ML-driven approaches in pharmaceutical R&D can help provide the necessary insights to guide sponsors with timely, evidence-based decision support based on predictions of clinical trial success within the context of an indication’s competitive landscape. 

Tangible benefits of AI/ML-driven solutions
Given the complexity of drug development programs and related costs and timelines, understanding the probability of technical and regulatory success for a specific intervention within an indication early on can impact clinical planning decisions, including which components to consider, such as inclusion/exclusion criteria, combination therapies, and geographies. 
 
For example, PD-(L)1 checkpoint inhibitors have transformed the oncology treatment paradigm with rapid expansion of treatments across multiple indications. As of 2021, a large majority of nearly 4,900 active clinical trials (83%) investigating PD-(L)1 inhibitors have been testing these therapies in combination with other immunotherapies, targeted therapies, chemotherapies and radiotherapies, and across a broad range of modalities. For those looking to make headway in this saturated marketplace, it is crucial to understand the intricate patterns within the latest developments to examine which combination therapies are showing promise. AI/ML solutions can parse the large volume of information to help oncology sponsors understand: 

• Where a sponsor’s potential PD-(L)1 development program stands relative to competitors.
• What potential combination therapies are worth considering. 
• Potential regulatory approval timing, assuming it’s likely, compared to others in the space. 

Optimizing curation of multiple data sources 
Pharmaceutical companies may have access to data from numerous trusted sources (e.g., trial registries, scientific literature, structural protein data archives, toxicology testing databases and their own proprietary data); however, it is time-consuming and costly to aggregate this information and extract meaningful insights. Instead, sponsors can use AI/ML-based solutions to parse millions of relevant data points from an extensive breadth of real-world data sets and rapidly deliver relevant insights for evaluation. 
 
AI/ML-powered tools can provide insight for hundreds of thousands of pathways, such as clinical development of a therapy targeting a disease starting from a first-in-human study and being abandoned or a therapy proceeding through clinical phases and obtaining regulatory approval, using hundreds of variables, including: 

• Intervention 
• Disease type
• Patient population 
• Previous trial outcomes
• Sponsor track record
• Chemical properties of compound 
• Toxicity profiles
• Relevant journal publications.

AI/ML-driven predictive insights 
From there, these numerous pathways can be used to train an AI/ML model to learn from clinical developments that have secured approval and those that were abandoned to predict how successful a new clinical strategy may be from both a technical and regulatory standpoint. 
 
As data sets vary, so will the predictive insights that are extracted from these AI/ML-based models. These tools can provide a calculated outlook in areas of importance for sponsors, including: 

• The competitive landscape for a specific mechanism of action or therapeutic area. 
• Indication-specific insights on pricing and reimbursement outcomes for disease-specific treatments at a country-specific level. 
• Mapping and scoring of all existing combination therapies. 
• Detecting promising assets by comparing early-phase trial data to approved treatment data. 
• Estimated dates of phase transition and regulatory approval, improving projections of development timelines for trial planning and design purposes. 
• Understanding a treatment’s technical and regulatory success and pricing potential to gauge options for licensing opportunities or acquisitions.  

Key components to elevating success  
A 2023 study published by Johns Hopkins University Press describes how individual experts are often uninformative when it comes to forecasting trial outcome and recruitment, and that trial results often surprise expert communities, who are frequently wrong about a treatment’s efficacy.¹ AI/ML-based approaches can provide objective and data-driven forecasts without personal biases to augment domain insights provided by clinical, medical and drug-development experts. This will accelerate their ability to analyze available data and provide recommendations in such areas as indication prioritization and clinical development planning. Underpinning these solutions with therapeutic, clinical development and data science expertise is key to gleaning deeper understanding about potential assets.
 
Pairing probabilities for technical and regulatory success with therapeutic and drug-development insights, such as mechanism of action, indications that were abandoned given lack of efficacy in previous trials and knowledge of historical success of treatments with similar molecular structures, can help guide sponsors’ clinical-trial design and planning decisions. 
 
AI/ML approaches can also quickly and efficiently identify key data within the ever-expanding scientific literature and can predict into which indications existing interventions can be repurposed, which is crucial for providing insight into clinical developments. Therapeutic experts who have clinical-trial experience in specific disease areas and intended patient population needs can then use these findings in the literature to provide contextual information about an investigational therapy that may play a role in its potential for success.
 
It is equally important to rely on experienced AI/ML experts who can help identify which types of granular data and relevant points of evaluation are needed to generate insights that effectively illuminate their level of opportunity and to understand how to optimize AI/ML-algorithms to drive these findings. 
 
Finessing market intelligence 
Using an AI/ML-based approach requires thoughtful planning of the key insights sponsors are looking to secure and which experts need to be involved in the process, because context is critical and the information gathered and evaluated can be optimized to best inform decisions. For example, leveraging AI/ML-driven insights to identify potential risks earlier in the process can enable sponsors to make trial design changes or reprioritize their investments. 
 
As sponsors consider the integration of AI/ML-driven approaches to better gauge the market and areas of opportunity, there is room to further fine-tune strategic direction in a highly competitive landscape by leveraging clinical-development insights to improve trial design and reduce participation barriers, managing trial complexity with effective design and execution of adaptive and novel trial designs, and increasing predictability and speed by applying real-world and commercial insights to clinical strategies and forecasting scenarios, and more.