The next level of competitive intelligence
Ketan Patel tells us why AI could be a game changer for business development and licensing teams
In this article Ketan Patel, Product Director - Portfolio, Licensing, and Clinical, Clarivate Analytics offers insight into three practical use cases for AI in the BD&L process.
To be effective in their role, business development and licensing professionals need to summarize and synthesize a vast amount of information from a variety of sources.
To understand the foundation of a valuation, teams must manually sift, sort and analyze drug pipeline information, sponsor data, financial information, deal data and clinical trial performance.
Not only is this data being generated daily, there is the added pressure of needing to rapidly and accurately evaluate the data before your competitors. Many teams often review the same company, deals or assets as their competition. If they can’t gather all of the required information in a timely manner, they will undoubtedly fall behind. For example, if you consider the process of valuing a small biotech with three core drugs in development, pulling together relevant information to understand the likely timelines for each drug’s approval will make a substantial difference in the valuation.
As this data must be pulled from disparate sources, there is also the issue of continuously monitoring useful information and ensuring this is being used within calculations. For business development and licensing, the core aim when gathering data is to use market information to make accurate predictions and forecasts.
In the quest to manage this process in a more efficient way, many have turned towards artificial intelligence (AI). By transforming the current manual evaluation process into an autonomous algorithm led process, business development and licensing teams will be able to work through more information and evaluate far more companies and drug assets. They can also improve their predictive ability by analyzing historical data and modelling several different success rates at once.
Currently, there are three key use cases where AI stands to make a sizeable impact on the business development and licensing process.
Being able to utilize historical data to identify drug development paths and time to approval
Across the public domain, there are a number of historical datasets available, including the clinical trials dataset and the FDA approvals dataset. This historical data can be extremely valuable when teams are seeking to identify the right path and time to approval for key drugs.
At Clarivate, we’ve used an AI tool to collate this public data and then add value to it by indexing it in specific ways. This enables the creation of a model which can then predict and identify the time to approval along with the probabilities for success.
Users can then value prospective companies or assets in the context of historical approvals and timelines. For example, when valuing a pipeline focused on Alzheimer’s you can view the historically low approval rate and review which key factors have led to approvals and consider whether the prospective pipeline is likely to succeed.
Or you may use an AI tool to analyze cycle times. If you can understand the historical average cycle times between different phase transitions, you can make a more accurate prediction for approval timelines.
Improved ability to forecast timelines and adjust plans
If you can compare the historical averages for certain drugs, you will be better able to forecast and also predict different timelines for similar products.
If a number of companies are also investing in the same area, you can likely predict when the first product will be approved, who will be next and so on. Then, you can look at the factors that directly impact these timelines.
Often, the first person to start a clinical trial will not be the first to gain approval. Several factors are at play. If you are able to understand the key factors, then you can adapt your current process to gain a competitive advantage.
For example, you could run a predictive model to see if using real world evidence during the activation phase of a clinical trial could increase your recruitment rate and thereby your time to market. If you have data which can show you which actions will generate the fastest approval results, you can adjust your internal plans to act on these forecasts and win the race to approval.
Ensure data driven evaluation of internal portfolios and external assets
Every company has its own mechanisms and methods to evaluate portfolios, comparatively analyze assets and spot market gaps.
These methods often come with inherent biases that can affect the valuation. These biases may come from simple misconceptions. For example, if your company has been successful in diabetes, there may be an assumption that these rates of approval will continue for future products. Or when evaluating a new indication, you may carry over information from another indication to guide your new area calculations. These instances are subtle and nuanced but can impact the decisions being made in a negative way.
By using an AI algorithm, you effectively have a data-driven independent advisor on your team.
When you don’t have that inherent bias, you can complete portfolio valuations and asset comparisons in a more accurate way.
Using this method you can also evaluate your competitors’ assets. An AI algorithm doesn’t just run across your own data, it gathers information across the entire drug development pipeline for any company and drug project. This allows you to predict the value of your competitor’s portfolios and address any internal gaps with suitable external assets.
It could also improve your out-licensing strategy by offering an accurate comparison between the assets you are looking to offload with what is already on the market. This will allow you to realize a stronger value for your products as you will better understand who stands to gain the most from the asset acquisition.
The data dilemma of artificial intelligence
For those seeking to explore the use of AI for business development and licensing, there is one key consideration to keep in mind: data.
Companies must understand both the sources of data and how frequently this data is updated. For an AI model to be effective, it needs access to all relevant sources of data and for these to be updated in a timely manner.
In the model we’ve built, it is updated nightly with all of the new data generated that day from different sources. This allows for any news that could impact a drug development program to be taken into consideration on a daily basis as the system makes predictions.
Quality of data is also a critical component. The phrase ‘garbage in, garbage out’ holds true, especially for models which have just been given large amounts of unstructured data. If you want to have confidence in the predictions an AI model makes, then you need to be sure that it is being fed and trained on quality data.
When technology is rapidly evolving, these tools should not be static. They should continuously evolve to take in new pieces of information, new types of milestones and new data types. The model should be upgraded so it can evolve alongside the industry. For example, the gene therapy market is currently progressing, but the data on approved gene therapies is limited. As the market matures, more data will emerge and companies need to be able to use this to make their predictions.
To hear directly from Ketan Patel about how to transform your business development and licensing capabilities, join us for an exclusive webinar.