Turning Uncertainty into Opportunity: Smarter Clinical Asset Evaluation Powered by Causal Machine Learning

Credit: WrightStudio | stock.adobe.com

It is an old story in the pharmaceutical world. Every year, countless drugs are developed at tremendous cost, yet only a fraction make it to market. However, as the saying goes, “in the middle of difficulty lies opportunity,” and behind this reality, there is a lucrative market for investing in or acquiring clinical assets that have not completed the development phase.

Savvy investors and pharmaceutical companies are increasingly recognizing the opportunity to “rescue” assets that otherwise would be neglected and repurpose them for success with a more targeted and systematic approach. Effective due diligence is key to identifying assets with hidden potential from those likely to fail, as well as identifying the development path required to take them past the finish line.

Investors commonly rely on tools for due diligence that focus on publicly available market data. However, these tools often fall short in delivering the analytical depth necessary to evaluate an asset’s true potential. While they offer a solid foundation for initial assessments, they lack the capacity to uncover deeper insights. As a result, due diligence efforts may remain surface-level, leaving critical data and subtle signals overlooked. This can limit an investor’s ability to gain a truly differentiated understanding of an asset’s trajectory and intrinsic value.

A more efficient and data-driven approach to clinical asset evaluation is needed. This is where causal machine learning (ML) can step in, helping investors dive deeper into trial data insights to evaluate the true potential of an asset and uncover new hidden opportunities.

Buying Low to Maximize ROI

Pharmaceutical companies often prematurely deprioritize or abandon clinical assets due to strategic shifts or reorganizations, offering them for sale. Leveraging ML to analyze the asset's raw data, investors can determine its potential value and pursue in-licensing opportunities at a relatively low cost, maximizing their return on investment by uncovering and developing undervalued opportunities.

The Case for Patient-Level Data

Many financial investors rarely have access to raw patient-level data during due diligence. Instead, they rely on qualitative benchmarks, aggregate data, and summarized graphs that may obscure critical information about a drug's performance and true potential.

Sophisticated biopharma investors recognize that accessing patient-level data in the diligence process can help identify drugs with a higher probability of success in future trials. However, lacking an effective tool to advise on a drug’s potential, they frequently also rely on qualitative benchmarks.

By leveraging advanced causal ML, these investors can analyze individual patient responses within clinical trials to identify patterns and irregularities that aggregate data might overlook. This patient-level view provides a differentiated point of view on a drug's efficacy and safety, which can help investors make more informed decisions and de-risk their investments.

Applying Advanced ML for Clinical Asset Due Diligence

Machine learning can help investors optimize the due diligence process in a number of ways:

1. Identifying Subgroup Populations Through Retrospective Analysis
   By leveraging causal ML techniques, investors can access the main factors driving heterogeneity, revealing the optimal subpopulation to maximize the treatment effect. For example, a drug that failed broadly may show efficacy when focusing on the right population. Causal ML can identify these subpopulations and offer the optimal clinical path leading to success.
2. Uncovering Hidden Safety Issues
   Safety signals that remained hidden in aggregate data can often emerge through ML's ability to analyze individual patient-level outcomes. Taking a proactive approach to safety can help investors identify potential risks earlier in the diligence process, reducing the likelihood of expensive setbacks.
3. Rescuing a Failed Asset
   ML can help investors assess the reasons behind a specific trial failure and determine whether the asset could succeed under different conditions. This may include a revised protocol, alternative dosing, or a new patient population.
4. Revealing Development Complexity and Costs
   Once a promising asset or population has been identified, ML simulation capabilities can uncover the optimal development strategy. This provides crucial insights into the costs and risks of the specific asset’s clinical development paths, which are inherent to making smart investment and acquisition decisions.
5. Validating Endpoints
   Endpoints are critical to trial success. ML can validate whether the selected endpoints are appropriate based on past trial data and emerging insights. This important step increases the likelihood that future trials will yield meaningful and actionable results.

Precise Path to Success

Beyond patient-level insights, ML accelerates the due diligence process by streamlining complex data analysis and leveraging predictive models. This reduces the reliance on manual, time-consuming methods. By predicting the likelihood of an asset's success or failure, ML tools minimize uncertainty, increase the chances of positive returns, and help identify hidden high-potential opportunities faster.

For financial players investing in clinical assets and pharma companies involved with in-licensing, ML offers an unparalleled level of precision to guide go/no-go investment decisions. Its ability to analyze raw patient data and predict outcomes creates new opportunities to improve current assets, rescue failed trials, explore alternative patient groups or indications, and increase the likelihood of trial success and eventual commercialization.

A Future of Opportunity Through Asset Evaluation

Applying ML tools for clinical asset due diligence provides a unique opportunity for investors to better evaluate the true potential of an investment and identify exciting opportunities in areas that would otherwise be missed. Those who embrace an ML approach can lead the way in bringing much-needed therapies to market, unleashing the potential of previously overlooked treatments.

As the drug discovery and development ecosystem continues to evolve, ML will play a growing role in helping investors make smarter data-driven decisions that accelerate drug development and commercialization which, in turn, maximize investors’ returns.

About the Author

Raviv Pryluk is co-founder and CEO of PhaseV, a pioneer in causal machine learning (ML) for clinical trial analysis and optimization. Prior to his tenure at PhaseV, he held integral roles in operations and data analytics at Immunai, a biotechnology company specializing in the mapping and reprogramming of immunology using machine learning and software engineering. Raviv holds a MSc in aerospace, aeronautical, and astronautical engineering from Technion - Israel Institute of Technology and a Ph.D. in neurobiology and neurosciences from the Weizmann Institute of Science in Rehovot, Israel.